ABIT IC7

[ThugsRook]

http://home.comcast.net/~d2rook/ic7/abitic7.gif

i have no predetermined format in mind. im just gonna post info about this board in no particular order and in no particular time intervals........

[ThugsRook]

for starters....

the IC7 can support 400 533 and 800 fsb cpus.
it can also support a P4M @ a 12x multiplier :deviltail

[TenEgg]

I thought you tested the IC7 already?

[ThugsRook]

yep, that was about 2 months ago.
the new bios files made this board worth taking a second look at.

:-)

[ThugsRook]

Dual Channel Shootout
TEST SYSTEM

P4 2.53B @ 3.4ghz 179fsb
Kingston HyperX 3000 2x512mb
Swiftech MCX4000 | 92mm fan | AS3
SBLive5.1 | WD800jb | Enermax 550w
GF4 ti4600 | det 43.03 | DX9a



TEST SPECS

ASUS P4G8X-D Granite Bay
bios x86 v1004 / cas2226 / Turbo / 1:1
http://home.comcast.net/~d2rook/ic7/p4g8xd.jpg (http://www.newegg.com/app/Showimage.asp?image=13-131-437-08.JPG/13-131-437-07.JPG/13-131-437-06.JPG/13-131-437-04.JPG/13-131-437-09.JPG)


ASUS P4P800 Springdale
bios v1008.003 / cas2226 / Turbo / MAM / 1:1
http://home.comcast.net/~d2rook/ic7/p4p800.jpg (http://www.newegg.com/app/Showimage.asp?image=13-131-459-04.JPG/13-131-459-03.JPG/13-131-459-01.JPG/13-131-459-02.JPG)


ABIT IC7 Canterwood
bios v14 / cas2226 / Auto / NBS533 / 1:1
http://home.comcast.net/~d2rook/ic7/IC7.jpg (http://www.newegg.com/app/Showimage.asp?image=13-127-153-09.JPG/13-127-153-10.jpg/13-127-153-07.JPG/13-127-153-08.jpg/13-127-153-11.JPG)



BENCHMARKS

> Aida32 MemBench < read - write
P4G = 4764 - 1192
P4P = 4996 - 1715
IC7 = 5006 - 1724


> 3DMark2001 <
P4G = 15782
P4P = 16063
IC7 = 16201


> UT2003 Demo < 1024x768, flyby - botmatch
P4G = 197.3 - 86.5
P4P = 198.5 - 89.1
IC7 = 198.9 - 90.2


> Quake III < 1152x864
P4G = 289.6
P4P = 289.7
IC7 = 291.1



* these improvements may look small, but on a gf4600 ~ they aint small ;)

:wave:

[pastorjay]

Nice THugs... I am going for round 2 on the Abit Canterwood as well... got me a "G" on the way to go with my new 2.8c! Good luck with the board... looking good so far!

PJ

[ThugsRook]

Bios Tweaking (v14)
* 179fsb 3.4ghz
* 3DMark2001


Cas Settings
2.5-3-3-7 = 15860
2.0-2-2-6 = 16088
2.0-2-2-5 = 16077


NorthBridge Strap
800 = 15749
667 = 16006
533 = 16088


AGP/PCI Lock
66/33 = 16088
72/36 = 16111
78/39 = 16164


Disable Unused PCI Clock
no = 16164
yes = 16198


System Bios Cachable
enabled = 16198
disabled = 16205


Video Bios Cachable
enabled = 16203
disabled = 16150


AGP Aperture
128mb = 16203
256mb = 16184


APIC Mode
disabled = 16184
1.1 = 16182
1.4 = 16203


Game Accelerator
Auto = 16204
Turbo = failed
Street Racer = failed
F1 = failed


Refresh Cycle Time
auto = 16203
normal = 16152
enhanced = 16205
strength = 16176
aggressive = failed


Read Delay Adjust
auto = 16186
11 = failed
8 = 15610
5 = 15950
4 = 16098
3 = 16171
2 = failed


Read Delay
auto = 16205
enabled = failed
disabled = 16160


Command Per Clock
auto = 16205
enabled = 16204
disabled = 16105


PCI Latency
32 clocks = 16187
64 clocks = 16189

:wave:

[pastorjay]

Whew! Now there is a bunch of numbers!!! Thanks Thugs! Alot of work there!

PJ

[ThugsRook]

Prime95 @ 179fsb 3.4ghz
* winbond hardware monitor
* room temp 76*F
* idle / load

cpu = 45*C / 63*C
system = 32*C / 34*C
pwm = 32.5*C / 43.5*C

vcore = 1.55v / 1.54v <----- set to 1.575v in bios
ddr = 2.7v / 2.7v <----- set to 2.7v in bios


:scratch: pwm = power management? the mosfets?

[ThugsRook]

according to my nVidia drivers...

AGP fastwrites is disabled but it also says its enabled :scratch:
(there is no option for this thru the bios)

http://home.comcast.net/~d2rook/ic7/cb40.gif


:wave:

[oldfart]

Game Accelerator
Auto = 16204
Turbo = failed
Street Racer = failed
F1 = failed
If you use PSB800, can you use a faster Game Accel? Is it faster? I can go all the way to F1 @ DDR412 1:1 2-2-2-5.

Was the fastwrites the same on the Asus? It does work on my 9700P, but does not make any performance difference.

[ThugsRook]

not yet.

fastwrites was enabled on the p4g8x and p4p800 and the drivers screen double confirmed it in both locations. however the IC7 cant seem to make up its mind ~ its unchecked but it still says its enabled. ill assume its enabled ;)

:wave:

[oldfart]

I know you try everything. That is one reason I'm glad you have the board! I dont have the patience to try ALL the options/combos like you do! I'm gonna have to try some of the ones from above and see how they do. I have the cachable stuff off and a 256M AGP app.

[LastRide]

Good stuff there chief.;)

[ThugsRook]

http://home.comcast.net/~d2rook/ic7/3502mhz.gif


:wave:

[ThugsRook]

IC7 vs P4P800 @ 184fsb 3.5ghz

TEST SYSTEM

P4 2.53B @ 3.5ghz 184fsb
Kingston HyperX 3000 2x512mb
GF4 ti4600 | det 43.03 | DX9a



TEST SPECS

ASUS P4P800 Springdale
bios v1008.003 / cas2226 / Turbo / MAM / 1:1

ABIT IC7 Canterwood
bios v14 / cas2226 / Auto / NBS533 / 1:1



BENCHMARKS

> Aida32 MemBench < read - write
P4P = 5133 - 1769
IC7 = 5149 - 1770


> 3DMark2001 <
P4P = 16255
IC7 = 16351


> UT2003 Demo < 1024x768, flyby - botmatch
P4P = 199.5 - 91.6
IC7 = 199.7 - 92.3


> Quake III < 1152x864
P4P = 291.7
IC7 = 293.3


> Sandra Unbuffered <
P4P = 3083 - 2933
IC7 = 3107 - 2974



:wave:

[Mikki]

The P4P800 is getting slaughtered....:eyepop:

Thugs, how does the IC7 handle oc'ing with the new bios? Any boot or reset issues?

I'm amazed at some of the differences in scores in the bios tweaks, pretty awesome...:)

[Duvie]

Thugs does those gaming modes work with more relaxed timings or are those just another feature of -200fsb chips???

Also is 3.5ghz the best you have gotten at 1.61v with that chip??? Just wondering if all these boards you have tested have all been similar in end clock speed...

[ThugsRook]

Mikki ~

the IC7 is a piece of cake.
no issues what so ever.

a couple things i really like about this board.....
* side mounted IDE ports
* room for ram swaps
* 4 phase power regulation
* plenty of OC options to play with
* 5 onboard fan headers
* well thought out item placements
* unstable settings wont even boot (hangs before post)

the only thing i dont like about the IC7 is the northbridge HSF.
id rather see a huge passive heatsink (one that can easily mount a typical 50mm fan if the user feels they need it)

again i used a large passive HS from an ASUS P4G8X on my IC7.

i still have alot of testing to do, but i like how easy this board is to work with so far.

:wave:

[ThugsRook]

Duvie ~

the gaming accellerators dont work well under 200fsb.
but on the bright side i can run the 533 northbridge strap where as the +200fsb ppl cant even dream of it.
i dont have any results showing which of the 2 is better tho (533nbs or turbo)
perhaps better memory will get me there ~ but the GA may just be for +200 systems.


how far the cpu will go depends on vcore voltage and how much im willing to give it. as we know well its hard to compare vcores and temps in a apples to apples comparison.

:-)

[ThugsRook]

in a few minutes im gonna test the Thermaltake NB HSF unit on this board.

http://www.thermaltake.com/images/products/chipset/productTiger1.jpg (http://www.thermaltake.com/products/chipset/tiger.htm)

the Thermaltake is extremely quiet but im not sure if it can hold my current temps ~ ill LYK :wave:

im also curious if the stock cooler will work ok w/o the fan?
(the fan unclips from the HS very easily)

[ThugsRook]

P95 passed @ 182fsb......

http://home.comcast.net/~d2rook/ic7/p95-182.gif

[ThugsRook]

well,
the Thermaltake NB HSF is working just fine.
no change in temps at all, at idle or under full load.

of course there is always a chance that the temps reported arent even a NB temp at all :rolleyes:

but at least the Thermaltake fan is silent.

:wave:

[oldfart]

One thing I like about Q3 is I get consistent results (and it doesn't take forever) ;). 3DMark scores will vary too much from one test run to the next to get accurate results when looking for small changes. UT2K3 seems pretty consistent. I'll run that one.

X-Isle is nV only?

[ThugsRook]

yep x-isle tech demo is nv only.
ut2k3 will serve you well but remember ~ a small gain in ut is big gain ;)

:wave:

[oldfart]

Link doesn't work...not a biggie.

BTW, in case you didn't know, you can use the F4 key to get a boot device menu like the <esc> on an Asus.

[ThugsRook]

F4 ~ TY :wave:

[Thor]

Damn nice. Looks like the Abit is winning the OCing crown so far. I'm getting pretty tired of these vcore swings on my Asus boards.

[oldfart]

VDimm mod (http://www.xtremesystems.org/forums/showthread.php?threadid=12914) for the truly hardcore.

[ThugsRook]

gotta love those easy to follow instructions..... :rolleyes:

interesting mod.
im not seeing any vdimm undervolting tho ~ im using 2.7v. (but i only need 2.5v)

:scratch:

[ThugsRook]

hmmm, this is very interesting :scratch:

ive been running some (ok, alot) of tests on the northbridge strap and gaming accellerator options.... and i think i have discovered what the gaming accellerator does :idea:

im working on the benchmarks now ~ so far it is panning out just as i suspected :deviltail

:wave:

[Duvie]

sounds interesting....I wondered why they all failed in your above test....Was it a gimmick, a trick (smoke and mirrors), or just getting instability out of bios options....

[pastorjay]

Ok, we are all waiting! SHow us the stuff man!!! :yummy:

[ThugsRook]

NorthBridge Strap Or Gaming Accelerator ???
first lets start off with a quick lesson on the NorthBridge Strap....
(bios v14)

Common Specs
179fsb | 1:1 (358ddr) | c2226
Game Accelerator = AUTO (disabled)

* 3DMark2001

NB/s 800 = 15749
NB/s 667 = 16006
NB/s 533 = 16088


533 is the fastest and 800 is the slowest.
with that knowledge lets move on to the gaming accellerator....


Common Specs
133fsb | 1:1 (266ddr) | c2226

* "AUTO" should actually be labeled "DISABLED"
* 3DMark2001 --- Aida32

NBs533, AUTO = 13844 --- 3713 / 1279 <
NBs533, Turbo = failed
NBs533, Racer = failed
NBs533, _F1_ = failed

NBs667, AUTO = 13709 --- 3553 / 1271
NBs667, Turbo = 13837 --- 3711 / 1280 <
NBs667, Racer = failed
NBs667, _F1_ = failed

NBs800, AUTO = 13461 --- 3371 / 1178
NBs800, Turbo = 13682 --- 3541 / 1186
NBs800, Racer = 13876 --- 3708 / 1606 <
NBs800, _F1_ = 13813 --- 3698 / 1302


im gonna try these same tests @ 167fsb......

[pastorjay]

Very Very Interesting... :wack:

[Duvie]

Interesting and maybe sad should be added.....The 800 strap needed these bios adjustments just to compete with the 533 at default auto.....The 667 never did....

What do you think this states about the current state of this product???

[ThugsRook]

i think the gaming accellerator changes the NBS.

turbo lowers NBS 1 notch lower then what its set at~
racer and F1 lowers NBS 2 notches lower then what its set at ;)
(making them closer to the performance of nbs533 AUTO)

"racer" obviously adds something else too ~ possibly 1T CMND :beer:

[shoman94]

:stop: j/k :smash:
It took a 533 CPU to figure this out......

Nice one THUGZ......:rock: :beer:

But this is definatly....:wack:

[ThugsRook]

haha ~ wanna see something funny?
lets add a +4:5 and a +3:4 mem ratios to the results.....
(+4:5 ratio is only available in NBs533 ~ +3:4 is only available in NBs400)

Common Specs
133fsb | c2226

NBs400, AUTO +3:4 (355ddr) = 14234 --- 3885 / 1652 <
NBs533, AUTO +4:5 (333ddr) = 14010 --- 3658 / 1579 <

NBs800, Racer 1:1 (266ddr) = 13876 --- 3708 / 1606
NBs533, AUTO 1:1 (266ddr) = 13844 --- 3713 / 1279
NBs667, Turbo 1:1 (266ddr) = 13837 --- 3711 / 1280
NBs800, _F1_ 1:1 (266ddr) = 13813 --- 3698 / 1302
NBs667, AUTO 1:1 (266ddr) = 13709 --- 3553 / 1271
NBs800, Turbo 1:1 (266ddr) = 13682 --- 3541 / 1186
NBs800, AUTO 1:1 (266ddr) = 13461 --- 3371 / 1178


the faster ddr still rules@ 133fsb...

:wave:

[ThugsRook]

NBs400: 3:4
NBs533: 1:1, 4:5
NBs667: 1:1, 5:4
NBs800: 1:1, 5:4, 3:2


NBs400: GAT does not work
NBs533: GAT does not work
NBs667: only GAT turbo works
NBs800: all GAT options work


Game Accellerator:
auto (disabled) = does nothing
turbo = lowers NBS 1 notch, must be in at least NBS667 mode to work
street racer = lowers NBS 2 notches, must be in NBS800 mode to work
f1 = same as racer

this is why NBS533 and 667 mode can be tricky. altho the IC7 offers you certain GAT options ~ there is no way for them to all work in all NBS modes. they only all work in NBS800 by design.

HTH :)

[pastorjay]

This has gt me itchin' to get my IC7-G here. SHould be here tomorrow, then I can run any tests you want me to run Thugs...

[ThugsRook]

thx PJ :-)
i think im gonna be keeping you busy for awhile ;)

im still working on these same tests @ 167fsb.
when im done that im gonna find out whats so "special" about racer mode :wave:

[ThugsRook]

NorthBridge Strap Or Gaming Accelerator ??? -part 2
Common Specs
167fsb | 1:1 (334ddr) | c2226

* 3DMark2001 --- Aida32

NBs533, AUTO = 15634 --- 4658 / 1603 <
NBs533, Turbo = failed
NBs533, Racer = failed
NBs533, _F1_ = failed

NBs667, AUTO = 15509 --- 4477 / 1611
NBs667, Turbo = 15580 --- 4658 / 1603 <
NBs667, Racer = failed
NBs667, _F1_ = failed

NBs800, AUTO = 15243 --- 4235 / 1477
NBs800, Turbo = 15491 --- 4449 / 1485
NBs800, Racer = 15637 --- 4648 / 2016 <
NBs800, _F1_ = 15582 --- 4652 / 1644


+4:5 mem ratio score
NBs533, AUTO, 4:5 (418ddr) = 15737 --- 4616 / 1973 <



* i need to find out whats so special about "racer" :scratch:

:wave:

[ThugsRook]

in case it wasnt clear ~ "racer" mode seems to make an additional change to the bios settings, besides lowering the NBs to 533....

Aida32 Membench

NBs533, AUTO = 4658 / 1603
NBs800, _F1_ = 4652 / 1644
NBs800, Racer = 4648 / 2016

looks like f1 on steroids right?
im gonna go thru the bios settings and see if i cant find out what that setting is....

aida32 is turning out tobe a VERY usefull tool! :beer:

[ThugsRook]

Common Specs
167fsb | c2226

hmmmm :scratch:

Aida32
NBs800, Racer, 1:1 (334ddr) = 4648 / 2016
NBs533, AUTO, 4:5 (418ddr) = 4616 / 1973

na, couldnt be! could it? :scratch:

[pastorjay]

Hehehe... I assume racer is also ONLY available at 1:1...
This is getting interesting!

[ThugsRook]

actually its the nbs800 that decides that...

NBs800: 1:1, 5:4, 3:2

because of my low fsb i dont think the negative ratios will effectively show me anything.

youll test them :na:

[pastorjay]

Yup, I 'll have her hooked up late tomorrow night... and will be able to run some then...

[ThugsRook]

here are my top 5 canidates.... (from my previous tests)

NBs400, AUTO, 3:4
NBs533, AUTO, 4:5
NBs800, Racer, 1:1
NBs533, AUTO, 1:1
NBs800, _F1_ , 1:1


ill LYK how far i can overclock each of them, their scores, and pick the winner.

(better ram might make this more interesting)

:wave:

[pastorjay]

Now this should be interesting!

My guess... 800 Racer 1:1

[SupermanCK]

can i say something...
I AM LOST!!! :eek: :eyepop: :smash: :eek2: :blush: :wack: :scratch:

[ThugsRook]

i dont blame ya SCK :wack:

Einstein once said, "you never truely understand something unless you can explain it to a child" :look:

once i post my final results of this "northbridge strap vs gaming accellerator" experiment, things will be clearer.


IC7 owners understand what im getting at here.
we have a lot of options in the bios, and since most users are using a 800fsb cpu they cant run most of the performance options available. so nobody really knew what these options can do.

since im using a 533fsb cpu i was able to get them all to run for me and show a good apples to apples comparison between them all. (at 133 & 167fsb)

:wave:

[ThugsRook]

i edited some of the previous posts to hopefully make things clearer.

for you non-ic7 owners out there,
the IC7 has alot of extra bios options to choose from besides the normal stuff.....

northbridge strap:
we have 4 options with 8 suboptions

Game Accellerator:
we have 4 options with 22 suboptions :look:

fun, fun, fun :wave:

[mysteryman]

Hey Thugsrook, first post here. Nice vbs board going here.
Wish AT would update the forum to something more current.
I like clicking that down arrow to read the latest posts...

My question for you is what the heck is NBs? What does it
actually do? And for an 800 MHz FSB cpu, can you set it to
anything?

I am an engineer, so I think I'll be able to understand...

To be honest, I am just doing my homework now so I can build
my first system when the prescott's arrive...

Thanks

[ThugsRook]

thx mysteryman :-)
...and welcome to BE.com :wave:

NBs (northbridge strap) is some sort of "ladder" that swings from 100 to 133 to 167 to 200fsb.
i think it gives the board makers the ability to "(de)tune" each fsb "zone" seperately.
and in typical OCing fashion we can now overclock them to a more aggressive "lower" strap for better performance.

but now there is also the "gaming accellerator" options. and it automaticly adjusts the NBs to its liking.

PJ will be repeating these same tests using a 800fsb cpu (@ 200+fsb) and he will post his results to confirm what he can and cant run.

HTH :-)

[mysteryman]

Thanks for the explanation! "lower" strap for better performance?
Typically better performance would be for a higher fsb. Strange.
Enjoyed reading yours, lastrides, pj's, oldfart etc. post for the past year.
You guys rock!

[ThugsRook]

yes, a lower NB strap is more aggressive and faster.
the only exception to this is "nbs800 racer" mode.

for 533 cpu OCers ~ the +4:5 mem ratio is the fastest.

HTH :)

[ThugsRook]

+mem ratio performance....
the +3:4 ratio (NBs400) died off quickly (@ 150fsb 400ddr) so its already out of the game.

the +4:5 ratio (NBs533) is very impressive but i need better ram to show its full potential :(

here's a quick comparison...


Common Specs
170fsb | c2226 | 2.8v

NBs800, Racer 1:1 (340ddr) = 15780
NBs533, AUTO 4:5 (425ddr) = 15982


:wave:

[LastRide]

I'll give that 4:5 NBs533 at 190-195FSB 2.4B a go soon as my IC7 comes in early next week.Since my OCZ Gold can handle about 520DDR it shouldn't be a problem:) .

[ThugsRook]

TY LastRide ~ that will interest me very much :-)

id go for a pair of OCZ but i want 2x512mb :look:

:wave:

[ThugsRook]

IC7 vs P4P800 @ 190fsb
Common Specs
3.61ghz | 190fsb 1:1 380ddr | c2226

* Aida32

P4P800 Turbo, MAM = 5298 / 1822
IC7 NBs800, Racer = 5321 / 2298

[pastorjay]

Wow! :eyepop:

[LastRide]

Wow nice scores.I can't believe your getting 3.6GHz from that chip.Thats a primo CPU.Would be nice to see that chip hooked up to a Prometia:eyepop:

[ThugsRook]

Northbridge Strap or Gaming Accellerator ??? -part3
COMMON SPECS
2.53b | 1.625v vcore | 1.65v vagp | 2.8v vdimm | cas2226


MAX FSB RESULTS

nbs800 1:1 Racer = 184fsb 368ddr
nbs800 1:1 _F1_ = 184fsb 368ddr
nbs533 1:1 Auto = 183fsb 366ddr
nbs533 4:5 Auto = 170fsb 425ddr



BENCHMARKS

> Aida32 MemBench < read / write
800 1:1 Racer = 5150 / 2228
800 1:1 _F1_ = 5137 / 1804
533 1:1 Auto = 5114 / 1757
533 4:5 Auto = 4696 / 2014


> 3DMark2001 <
800 1:1 Racer = 16421
533 1:1 Auto = 16351
800 1:1 _F1_ = 16338
533 4:5 Auto = 15881


> UT2003 Demo < 1024x768, flyby - botmatch
800 1:1 Racer = 199.7 - 92.6
800 1:1 _F1_ = 199.6 - 92.2
533 1:1 Auto = 199.6 - 91.9
533 4:5 Auto = 197.4 - 87.0


> Quake III < 1152x864
800 1:1 Racer = 293.6
800 1:1 _F1_ = 293.0
533 1:1 Auto = 292.8
533 4:5 Auto = 288.5



* nbs800 1:1 Racer is the fastest option for me right now, but with better ram the 4:5 ratio would be the fastest if it could hold c222.

:wave:

[ThugsRook]

ive got some KHX3500 on the way :-)

[LastRide]

:rock:

[Mikki]

Thugsrook, darn nice review, darn nice results, darn nice info. And thanks for making the effort to explain it to us non-owners....;):)

[ThugsRook]

hehe~ yea sorry about that Mikki! :beat:

if anybody has any questions, please ask!


here's a quick clear summary....

for 800fsb cpu OCing...
very best = nbs800 racer (1:1)

for 533fsb cpu OCing...
best = nbs800 racer (1:1)
very best = nbs533 auto (4:5)

of course this is subject to having the ability to run those settings/speeds

:wave:

[SupermanCK]

Well I might have fixed my problem.. Read on another forum that you should disable the last two memory settings in the game accelerator menu, ie Read Delay Adjust and Command Per Clock.. Before I did this, I could not get any higher than 260 FSB stable using 2XCorsair 512MB PC3500C2.. Once I changed the two settings to disable, default is auto, I can now run at 283 FSB prime95 and 3DMark2001SE stable.. So if you're having issues like I was, give it a shot.. - by legendma

[ThugsRook]

i have no problems with those 2 options being on auto (enabled)
turning them off slows me down.

might just be an XMS issue :scratch:



from my bios tweaking guide...

Read Delay
auto = 16205
enabled = failed
disabled = 16160

Command Per Clock
auto = 16205
enabled = 16204
disabled = 16105

[ErikaeanLogic]

if anybody has any questions, please ask!

I have a question: do MAM and/or GAT work at 5:4 ratio on either the IC7 or IS7? I haven't been able to determine this (one way or the other) from all the reviews I've read. My P4P800 has struck out with me (stopped posting about all the flakes this board has, THUGS has addressed many of them articulately;)) and I'd like to go with Abit's superior bios and power management (Rubicon caps). Guess my days of staunch support of Asus are gone:rolleyes: .

[ThugsRook]

PAT (canterwood) is supposed to work (by default) in all ratios.

its hard to tell exactly what its worth cause we cant turn it off!
however it is most effective in 1:1 mode.

the IC7 is an awesome board and is definatly THE BEST BOARD EVER for 533 cpu OCing. (i cant comment on 800 cpu)

:wave:

[oldfart]

Originally posted by ErikaeanLogic


I have a question: do MAM and/or GAT work at 5:4 ratio on either the IC7 or IS7?
Yes, it works (quite well) on the IC7. You have to be running BIOS ver 14. I dont have an IS7, so cant comment on that one.

[ErikaeanLogic]

Thanks ya oldfart:D

How about you IS7 owners out there? I want to hit the "submit order" button on a cart I've got queued with an IS7 in it but need your go-ahead:).

Thanks,

Erik

[Duvie]

What is the going price difference between the two??? Check out prometheus's review of the board in the motherboard section....Aandtech also has a nice roundup that shows the abit is quite a nice board and maybe the best of the i865 boards...

[pastorjay]

Well, let me just say, I have had them both (as well as a few other Canterwood/Springdale boards), and I am back on an IC7 (got the G this time around). The IS7 was just not stable for me. COuld never really get her dialed in. Could have been the ram timings, could have been impatience on my part.

In my opinion, the IC7 has been the easiest to work with so far. And I love the SoftMenu bios

[Duvie]

That is a raving endorsement!!! What will the "G" model give you pastorJay???

[Tony]

Thugs
IC7 is canterwood..has PAT
P4P800 is springdale...has PAT of a sort

My AX4C max is faster than my P4P800 upto 260fsb but won't clock much more over that.If the asus had the missing I/O chip that the review boards had im sure with the 533 strap set the boards would be just as fast..thing is they don't.

The one to look out for it the rev2 DFI coming soon.Also the Ax4CmaxII looks good but may only be an asian release :(

[pastorjay]

Well, I am slowly working on getting these Game Accelorator settings to work. I have some timing issues to work out with my OCZ 3500EL, I can run 250fsb, 5:4 and 235fsb 1:1 with no problems, but I cannot run any GA on these settings. I went 1:1 and the best I can do so far is run Street Racer at 215fsb, 1:1. But it is incredible the scores I am getting! Here is a sample:

http://www.fox302.com/userdata/pastorjay/files/IC7G/aidaread.JPG

http://www.fox302.com/userdata/pastorjay/files/IC7G/aidawrite.JPG

PJ

[ThugsRook]

W :eek2: W

[oldfart]

Thats with the 2.8? 250/235 FSB = 3.5/3.29 GHz. Nice! Its hard to run the aggressive GA settings @ high FSB. I can use all of them @ stock 200 FSB. I didn't try them @ my max 1:1 FSB of 213 (ram limitation). With my FSB @ 257 5:4 (CPU limit), I can only run auto.

[pastorjay]

I actually did not make myself rel clear... that is only at 215 fsb and 1:1! I can't imagine what the scores will be like when I get my OCZ GOld on Monday and can go to 250 1:1 (hopefully).

I'll have another screen in a moment...

PJ



Originally posted by oldfart
Thats with the 2.8? 250/235 FSB = 3.5/3.29 GHz. Nice! Its hard to run the aggressive GA settings @ high FSB. I can use all of them @ stock 200 FSB. I didn't try them @ my max 1:1 FSB of 213 (ram limitation). With my FSB @ 257 5:4 (CPU limit), I can only run auto.

[pastorjay]

Here is Sandra unbuffered!
215 fsb 1:1
http://www.fox302.com/userdata/pastorjay/files/IC7G/sandraunbuf215.JPG

[oldfart]

Yeah, I understood the Aida chart was the 215.

How about some benches of some real apps?? Or is that gonna happen after you get the new mem?

[pastorjay]

I'll likely do a few now and after the new mem... anything in particular you want to see?

[oldfart]

Honestly...

ALMOST ANYTHING other than SiSoft and Aida!! Something that you actually USE A PC FOR.

I dont know about you, but I didn't build and dont own a PC just for running synthetic benches. I play games (not as much as I used to). I do some video encoding with TMPGEnc and Pinnacle Studio 8, MP3 encoding, as well as the usual office/web stuff.

Games benches, video or audio encoding, archiving (Winzip/WinRAR) would be cool.

[LastRide]

Its funny how the 2.6,2.8,3.0C chips smoke the 2.4C even if your clocking the 2.4C higher.Seems like FSB has no impact.I am lost on why results differ so much.I have noticed this with oldfarts 2.6C,bigtoes 3.0C and now PJ's 2.8C.It doesn't matter how far you clock a 2,4C,any other chip will beat it with less overclocking.Even running 2.4C with a higher DDR coupled with a higher overclock will not match it.:scratch:.

[Duvie]

Hmmm that is perplexing!!!! Wonder why??? Board problems??? I noticed with thugs testing that sometimes the 800fsb oevr the 533fsb was worse. It is the chipset just not refined???

Thugs always specualted in the past that when the chip required more voltage there was a diminishing return in performance. Kinda like the chip misfired a bit and thus slowed it done slightly.....I don't know are you comparing say a 32.4@3.4ghz w/ 1.65+v to a 3.0c at 3.4ghz with default???

I personally think there may be something whacked in the actually i865/i875 chipset that starts becoming noticeable the more you oc the fsb....

[ErikaeanLogic]

Originally posted by pastorjay
Well, let me just say, I have had them both (as well as a few other Canterwood/Springdale boards), and I am back on an IC7 (got the G this time around). The IS7 was just not stable for me. COuld never really get her dialed in. Could have been the ram timings, could have been impatience on my part.

In my opinion, the IC7 has been the easiest to work with so far. And I love the SoftMenu bios

good enough for me, I'm in;)

thanks, guys, what would I do without you?:wave:

[oldfart]

Originally posted by LastRide
Its funny how the 2.6,2.8,3.0C chips smoke the 2.4C even if your clocking the 2.4C higher.Seems like FSB has no impact.I am lost on why results differ so much.I have noticed this with oldfarts 2.6C,bigtoes 3.0C and now PJ's 2.8C.It doesn't matter how far you clock a 2,4C,any other chip will beat it with less overclocking.Even running 2.4C with a higher DDR coupled with a higher overclock will not match it.:scratch:.
Odd. Anyone have a 2.4C and a 2.6C or 2.8C "in stock" and can do few benches? The only explanation would be that there is a point where too much FSB starts to degrade performance because the chipset is pushed too hard. A 2.4C Vs 2.6C or 2.8C @ ~ 3.3 - 3.4 GHz on the same rig would be an interesting test,

[ErikaeanLogic]

Will do, what do you want to see and at what MHz? I think I can do 3.3GHz (175fsb) with pretty tight timings with my 2.4C, maybe more. Will 3.3GHz do? Just lmk which benches you want to see;).

[oldfart]

Actual speed is not all that important as long as they both run the same CPU speed. Anywhere ~ 3.3 or so would be fine.

Benches....the usual game benches such as UT2K3, Serious Sam, Q3, 3DMark, etc (whatever you have). Video and audio encoding would be cool as well as Winzip or WinRAR archiving.

Should be a valuable test!

[ErikaeanLogic]

Will do. I should still have results from my 2.6C laying around here somewhere;). I'll get on it tonight.

[pastorjay]

Here is my first run of 3dM 2001...

Settings: 215fsb, 1:1 mem, Street Racer

http://www.fox302.com/userdata/pastorjay/files/IC7G/2153d01.JPG

Not my highest ever, but very good I think for 215fsb and no overclock on the card itself

[ThugsRook]

found a nice tweak in the bios when using a +4:5 ratio....

Common Specs
2.53b @ 3.23ghz 170fsb 425ddr
cas2227 | 2.8v dimm


Read Delay Adjustment

* 3DMark2k1 --- Aida32

Auto = 15879 --- 4696 / 2014
Disabled = 15861 --- 4673 / 2007
Enabled = 15982 --- 4892 / 2012

[SupermanCK]

what about disable? or is it auto = disable?

[SupermanCK]

oh yeah...i am starting to like this board more and more...just one thing tho is that there is no onboard LAN...i really have no use for the firewire ports... only if they offer onboard LAN version without going up to the G version, which cost $50 more...

[ThugsRook]

SCK ~ added disabled results for ya

i use a PCI NIC

:)

[pastorjay]

http://www.fox302.com/userdata/pastorjay/files/IC7G/ut2003215.JPG

Ok, how is this...

[pastorjay]

edited the UT scores

[SupermanCK]

i guess auto = disabled...
ty for adding disable thugs...
oh yeah btw pj...why are your pictures showing so much white area?
and very nice results...

[pastorjay]

not sure about the white area... I am using paint in XP and I likely did not crop them right... will work on that... :blush:

[oldfart]

Originally posted by SupermanCK
oh yeah...i am starting to like this board more and more...just one thing tho is that there is no onboard LAN...i really have no use for the firewire ports... only if they offer onboard LAN version without going up to the G version, which cost $50 more...
Lan is no big deal really. PCI nics are cheap enough. I like the firewire since I have a DV camcorder.

PJ, looking good so far (ya gotta crop those pics). Paint does not allow crop, but I think you can cut a piece out and paste into a new doc??

[pointreyes]

Originally posted by pastorjay
not sure about the white area... I am using paint in XP and I likely did not crop them right... will work on that... :blush:

You don't have to crop them. To capture the active window-press <Alt>-<Print Screen> at the same time will capture the active window for ya. :) The active window is the one with a darker title bar compared to the rest of the windows on the desktop.

Note: If you want to see if the screen is captured correctly you can either paste into wordpad or use the clipboard itself. You can get to the clipboard from the Run Prompt using the command: clipbrd.

[ErikaeanLogic]

right, but make sure you re-size the Paint canvas before you paste the image (make it very small);)

[SupermanCK]

man...i am telling yall....u learn something new everyday in this forum....:wave: :p :D

[pastorjay]

That's cause this place is the best!

[SupermanCK]

Originally posted by oldfart
Lan is no big deal really. PCI nics are cheap enough. I like the firewire since I have a DV camcorder.

PJ, looking good so far (ya gotta crop those pics). Paint does not allow crop, but I think you can cut a piece out and paste into a new doc??

i guess ur right...i mean i have a nic sitting around somewhere...but i think it's ISA interface...LOL

[pastorjay]

This may seem real crazy, but I have used both ad-on Nics and onboard for awhile, and it just seems that the onboard is a little faster... I really notice it with the Gigabit Lans... I cannot tell a difference with CSA though..

Wish there was some bench to test it all...

[SupermanCK]

maybe you can set up some sort of transfer and time it manually...like a video file (700mb)...

[shoman94]

PJ...what can you do with that 2.8C...I'm curious.....
Also what sSpec it your cpu?

SHO

[pastorjay]

L317A819
sl6wj

[ErikaeanLogic]

Originally posted by pastorjay
That's cause this place is the best!

. . .erm, but you didn't fix the previous attachments yet, PJ:na:

[ThugsRook]

bios v14 vs v15
* P4 2.53B @ 3.47ghz 182fsb 364dcddr
* NBs800 Street Racer


> 3DMark2001 <
v14 = 16329
v15 = 16360


> X-Isle Tech Demo <
v14 = 142.3
v15 = 142.6


> UT2003 Demo < 1024x768, flyby --- botmatch
v14 = 199.3 --- 91.9
v15 = 199.5 --- 91.8


> Quake III v1.32 < 1152x864
v14 = 293.0
v15 = 293.0


> Aida32 <
v14 = 5089 / 2210
v15 = 5108 / 2241

[pastorjay]

Ok, here is my latest.

250fsb 5:4 mem @ 2.5-3-3-6

http://www.fox302.com/userdata/pastorjay/files/IC7G/25054.jpg

[ErikaeanLogic]

wow, PJ, impressive scores!:rock:

As I promised on Friday, I ran a series of benches on my 2.4C at 3350MHz (compared to 3358MHz results with a 2.6C) in order to see what the effects of fsb are using high-fsb P4 overclocking. The mem timings in both cases are 2-2-2-5 at 5:4 (was using Corsair XMS3200C2 with the 2.6C, HyperX3500 with the 2.4C), the bios is 1008.004, and all other components are the same (WD800JB, SB Live!5.1, onboard LAN, Radeon 9700 Pro).

P4 2.4C @3350MHz/279fsb/223 (results in plain text)
P4 2.6C @3358MHz/258fsb/206 (results in bold text)

Without further ado. . .

SuperPi

41
41

UT Bench (1024 x 768)

214.4/94.3
214.2/94.8

3DMark2001 (build 330)

17,136
17, 238

Code Creatures

38.7
38.7

Q3 Bench[/b] (1280 x 1024, settings max'ed)

296.1
296.7

[u]Sandra Memory Bandwidth (buffered)

5574/5649
6058/5987

Sandra Memory Bandwidth (unbuffered)

2955/3018
3113/3183

Please keep in mind that the 2.6C benches were done just as the 1008.004 bios came out (2 weeks ago?) and the 2.4C benchies were done yesterday; alot of "shifting" of settings has taken place so the results are tentative, at best.

[mysteryman]

How do you switch the cpu out? Do you have to reapply as-3?

curious... sounds like alot of work...

[ErikaeanLogic]

LOL, well it is "alot of work" but what can I do? I'm addicted!:yummy:

Simply remove the heatsink, lift the ZIF socket lever, remove cpu, apply thin layer of thermal paste (I smear off the excess with my fingertip) to new cpu, insert new cpu into socket, close lever, reapply heatsink. Repeat:wack:

You'll know you're into it when you start timing how long this process takes in order to distract your brain away from the fact that you do it so (too?!) much.:wave:

[oldfart]

Originally posted by pastorjay
Ok, here is my latest.

250fsb 5:4 mem @ 2.5-3-3-6

http://www.fox302.com/userdata/pastorjay/files/IC7G/25054.jpg
Dont think I've seen anyone hit 100 in Botmatch yet.

You are close!!

[pastorjay]

Hopefully my GOLD will help me out to get 100. Also, I have yet to overclock the vid card...

[ErikaeanLogic]

you o/c that card, PJ, and 100 will be yours for sure;)!

[mysteryman]

my hats off to you mate! have alot of respect for you all.
:beer:

edit: you got a removable mb tray? i want that nice antec 1080amg, but it don't got that.



Originally posted by ErikaeanLogic
LOL, well it is "alot of work" but what can I do? I'm addicted!:yummy:

Simply remove the heatsink, lift the ZIF socket lever, remove cpu, apply thin layer of thermal paste (I smear off the excess with my fingertip) to new cpu, insert new cpu into socket, close lever, reapply heatsink. Repeat:wack:

You'll know you're into it when you start timing how long this process takes in order to distract your brain away from the fact that you do it so (too?!) much.:wave:

[kevvv]

I just went through two large coffee's reading this post...great stuff....

A question to all, since you guy have a lot of expereince with the two boards I'm looking at:

How do you think the IC7 would be with a 2.4c as opposed to the P4p800?

Also is the only difference b/w the Ic7-g and IC7 the GigLan?

And is PAT on the Ic7? The spec page I looked at at mwave only made it seem like PAT was on the IC7-G


Thanks all and keep up the great work!

[ErikaeanLogic]

hi, kevvv

Whether the P4P800 is better than the IC7-G for a 2.4C is a question I will be able to answer on Thursday, which is when my IC7-G arrives.

The primary differences between the IC7-G and non-G are an extra SATA RAID chip, the Gigabit LAN, and that the LAN runs over CSA (rather than the PCI bus).

HTH:),

Erik

[kevvv]

Thanks for the reply! looking forward to see what your results are. How far did you take it with the p4p800?

[ErikaeanLogic]

I run it all day long at 279fsb, but have run benchmarks at 292fsb although not totally stable. The Northbridge definitely needs an active cooling solution, as it gets mighty warm up there;).

[Mikki]

Whoa!! Did this thread get hijaked or what? :look:

[pastorjay]

Yup, I probably should have started my own...

Sorry Thugs!

PJ

[ErikaeanLogic]

yep, I'm guilty of P4P-ing on your thread:blush:

I'm sorry 'bout that, y'all:)



THUGS, what've you been up to with the IC7 lately? Haven't heard anything out of you in a while. . .soldering the Vdimm mod?:na:

[LastRide]

I'll have take some of the blame too.I did delete three or four postings here on my P4P800.Since I got my IC7 today the P4P800 is no more.:chit:

[ThugsRook]

Originally posted by Mikki
Whoa!! Did this thread get hijaked or what? :look:
these guys just cant stand it when i have nothing new to talk about ;)

got my KHX3500 today ~ expect a review. (on the IC7)

im also working on modding my vid card for a bit more OC out of it.
ive got a few different ideas in mind~
1) a socket7 cooler
2) modified stock cooler with a bigger fan
3) a fan card with 2x92mm fans :deviltail

that will also get its own thread if its worth anything.

im also working on TMPGenc to see if its worth anything as a benchmark. (built in timer)

:wave:

[kevvv]

sorry for hi-jacking.

I've got to do something while waiting for your next move... :D

[Mikki]

hehe....when the cat's away.....;)

I'm interested in the fan card, I've been thinking of that myself....it may have to wait tho, I just got a new video card to play with.....:D :deviltail

[oldfart]

Originally posted by ThugsRook
im also working on TMPGenc to see if its worth anything as a benchmark. (built in timer)
It is. I've used it many times. The only thing is, it's good for benching changes to your own system, not really for people to compare to each other. You would have to use the same source video files to do that.

A real good CPU power test is MPEG2 SVCD with 2 pass VBR highest quality. Lots of calculations going on.

Another thing I like about TMPGEnc as a bench is it's consistent. Its like Q3A. You can run the same bench several times and get the same results each time. Not like Sandra and 3Dmark where you get a different result each run.

[ThugsRook]

the HyperX 3500 sucked ~ no review :beat:

typical 5ns garbage :smash:

[pastorjay]

Blah!!! Hate to hear that....

Got my OCZ Gold today Thugs....

Good stuff... will post more tomorrow

[pastorjay]

By the way... here is my latest:

250fsb mem 1:1, 2.5-3-3-7, Turbo

http://www.fox302.com/userdata/pastorjay/files/IC7G/UTrecord.jpg

OVER 100! Whew hew!!!

[oldfart]

Very nice! How much Vcore do you need to get to 3.5 GHz?

[pastorjay]

I am at 1.575 atm

[oldfart]

mmm Nice Chip!:beer:

[SupermanCK]

woohoo..someone finally broke 100fps in botmatch
congrats

[shoman94]

Originally posted by pastorjay
I am at 1.575 atm

Is that prime stable?

[pastorjay]

3d stable... Prime is in Progress

[LastRide]

My IC7 has been a nightmare for me.Playing with it all last night.It won't do the same settings that the P4P800 did.It doesn't like my OCZ gold sticks that run great on the Asus.Memory goes back to by speed with low memory settings that should boot no problem.I tried both the 2.4B and 2.4C in it.It likes to hang on PCI device listing on boot.Instead of blue screen of death I get black screen of death..LOL.Sometimes I can boot at say 280 FSB,but if I reboot screen goes black and shuts down.

[SupermanCK]

remember what thugs said...the ic7 has a much more aggressive memory setting than the p4p800...so you might have to relax ur timing to get to the same spec...

[ThugsRook]

Originally posted by LastRide
My IC7 has been a nightmare for me.Playing with it all last night.It won't do the same settings that the P4P800 did.It doesn't like my OCZ gold sticks that run great on the Asus.Memory goes back to by speed with low memory settings that should boot no problem.I tried both the 2.4B and 2.4C in it.It likes to hang on PCI device listing on boot.Instead of blue screen of death I get black screen of death..LOL.Sometimes I can boot at say 280 FSB,but if I reboot screen goes black and shuts down.

i thought OCZ tests their ram on a IC7? ....duh!
OCZ doesnt like tight timings ~ and if you want fast memory performance you need to run 222 <period>

case and point: my membenches.

why did the KHX3500 suck?
1) cause it cant run 222 as far as the KHX3000
2) 5ns ram is just plain SLOWER in ALL BENCHMARKS! <period>

* ive tested enough ram to claim this as FACT.

:wave:

[oldfart]

Is the KHX or OCZ Winbond CH5 ram? The IC7 is known to HATE memory based on Winbond CH5, but BH5 is ok. The main problem I've seen is the inability to run 5:4 ratio with it.

There are a few threads on this subject on the Abit furms. Here is one of them (http://forum.abit-usa.com/showthread.php?s=&threadid=13877)

[ErikaeanLogic]

Originally posted by ThugsRook
i thought OCZ tests their ram on a IC7? ....duh!
OCZ doesnt like tight timings ~ and if you want fast memory performance you need to run 222 <period>

case and point: my membenches.

why did the KHX3500 suck?
1) cause it cant run 222 as far as the KHX3000
2) 5ns ram is just plain SLOWER in ALL BENCHMARKS! <period>

* ive tested enough ram to claim this as FACT.

:wave:

hmm, my 2 x 512 KHX3500 HyperX does 223fsb at 2-2-2-6 on my P4P800, I wonder if it'll do the same on the IC7-G (one more day, baby:)!)

[ThugsRook]

my KHX3500 was able to go 438dcddr (+4:5) cas2327 @ 2.8v.

the problem is that the loss of the 2 fast cas settings makes performance slower and not worth running the 4:5 at all.

even in 1:1 street racer mode ~ the KHX3000 toasted the 3500.
benches coming soon.....

cas settings are more important right now

[ThugsRook]

KHX3000 vs KHX3500
* P4 2.53B @ 3.47ghz 182fsb 364dcddr
* 1:1 nbs800 street racer c2226 @ 2.8v

KHX3000 6ns pc2700
KHX3500 5ns pc3200


> 3DMark2001 <
KHX3000 = 16375
KHX3500 = 16292


> X-Isle Tech Demo < 1024x768
KHX3000 = 142.8
KHX3500 = 142.3


> UT2003 Demo < 1024x768, flyby --- botmatch
KHX3000 = 199.4 --- 91.7
KHX3500 = 199.0 --- 90.5


> Quake III v1.32 < 1152x864
KHX3000 = 318.3
KHX3500 = 316.3


> Aida32 <
KHX3000 = 5090 / 2208
KHX3500 = 5083 / 1297 <--- :scratch:



* tweaks were attemped on the 3500 but to no avail.
* the 3500 wasnt fast enough to make +4:5 worth running
* the 3500 got to 175fsb 438dcddr (+4:5) c2327 2.8v
* the 6ns 3000 is faster in all tests.

this isnt anything new. ive run these same 5ns vs 6ns tests on every moboard ive owned for the last year ....6ns is ALWAYS faster.

:wave:

[ErikaeanLogic]

wow, what a difference!:eyepop:

I'll test them out, too, since I still have my 6ns XMS;)

[kevvv]

how do you all absorb all this information!!

I think I've read about five differnt types/brands of memory..

more power to you all!!

Keep up the good work!

[LastRide]

Customer support told me that they test the ram on an IC7 so why it runs like crap on mine I have no idea.Plain and simple the OCZ must not be good on an IC7.Then again,you see bigtoe and PJ run it with no problems.Oh well,I'll pay around some more and see what happens. :wack:.By the way,your really squeezing that board Thugs.:)

[shoman94]

Originally posted by ThugsRook
gettin a little more mhz out of my vid card ~ i modded the stock HS, replaced its fan, used ASC.

http://home.comcast.net/~d2rook/ic7/16548.gif

:wave:

WHat are you at for Ocing that card?

My ti4200 is at 295/670 here is my result....I'm wondering about the minor gap....
Detailed Test Results
3DMark Score 16188 3D marks

[ThugsRook]

that score was obtained @ 325/725.

:)

[Mikki]

:eyepop: :look: :rock: :wave:

[Thor]

Well, after reading all these IC7 threads on the 'net, I decided to go for one, as prices are pretty reasonable for the non-G version. Going to see if my 2.4C has anymore FSB left in it. :D

Thugs, thanks for the info and testing you and everyone else contributed to this thread. It will help out greatly for an Abit newb such as myself. :p

[kevvv]

what memory are you going to use with it? seems like OCZ is a crapshoot...

[ErikaeanLogic]

hey THUGS, you still running the v1.4 bios or v1.5?

oh, and why?;)

thanks:)!

[ThugsRook]

bios v1.5 ~ better performance and its more stable.

:-)

[Thor]

Originally posted by kevvv
what memory are you going to use with it? seems like OCZ is a crapshoot...

Corsair TwinX3200LLs.

[ThugsRook]

325x750 (no artifacts)
http://home.comcast.net/~d2rook/ic7/16947.gif

im so close! :wave:

[LastRide]

Close but no cigar.:na:

[ThugsRook]

i think it may be time for me to remove the ICH (heatspreader) :deviltail

[LastRide]

:eek2: gotta be another alternative.


Originally posted by ThugsRook
i think it may be time for me to remove the ICH (heatspreader) :deviltail

[ThugsRook]

na, ive been putting it off long enough.

ive got a 2.4b C1 comin this week so ill have plenty of time to (carefully) hack away at my 2.53b while im testing it.

the man (SinfulWeeper) got an extra 7fsb 133mhz from his 2.53b :eyepop:

:-)

[LastRide]

I know your Nvidia loyal but I think you should look into getting a 9700 Pro.The 4600 is still a pretty good card but its time to move on to better things.I mostly owned Nvidia myself but took the plunge.I was worried about the driver issues related with ATI.Their drivers are better than they used to be and I have had no problems myself.You can pick one up for a song and a dance in any forum.The 9800 Pro isn't that much more performance gain.Then you will be talking 20,000 :rock:,well maybe 19,000

[ThugsRook]

i really dont wanna buy a new card til Doom III.

at that time we will all be benching D3 and ill know what card to get.

:)

[Silencer]

Greetings
My first post here, but i was reading your posts every day for the last month while changing an Albatron 965 pe pro for a p4p and now the ic7. You`re really great guys.
For people with IC7 and OCZ Gold, is it posible that we are limited to run memory timmings "by speed" because the limited ic7 ddr voltage (2.8max)? OCZ says that the ram runs at 2.9v.
I cant breack 287fsb, at 290fsb it crashes windows.
I disabled tha last two game settings and hyperthreading to stay at 3.43ghz stable.
Sorry my english is not really good.

[OCZguy]

Try disabling cpc to squeze some more out of it

[ThugsRook]

the OCZ guy ~ hi! :-)

i think he may have already done that (the last 2 gaming accelerator suboptions)


the 4 suboptions are:

Refresh Cycle Time ~ try normal

Read Delay Adjust ~ try 8 (or the highest you can go)

Read Delay ~ try disabled

Command Per Clock ~ try disabled

those settings should be the slowest ones possible.
HTH :wave:

[LastRide]

You should be happy with 287.I can do 280 FSB 5:4 with my 2.4C and IC7 and OCZ 3700 Gold that I got to 292 FSB on the P4P800,not stable though.My ram runs 5:4.I run 280 FSB 2/3/3/6 448 DDR.Game Acelleration auto,auto,disabled,disabled.I have one stick that can do 500 DDR 2.5/3/3/7 memtest86 pass,all tests cache on,5 passes.My replacement stick (original speed chip broke and the palce I bought it from had to send me a stick twice cause the first one was a stinker) would get 2 errors on test 4 same settings 2.8 vdimm.My sticks aren't matched which kinda ticks me off.I tested them individually and windows was a little unstable with the stick with 2 errors at those setting and the other one wasn't.Being limited by 2.8 vdimm on the IC7 and sticks are not matched,to run 500 DDR I will have to use 2.5/3/4/7 cause of the one stick.

**Installed the new Omega 3.5 drivers 200 point increase 20,727 3DMark2001SE** 1024 x 768 :rock:

[Silencer]

LastRide, do you have a 9700 or 9800pro?

[kevvv]

Thugs,

Have you done a comparision b/w KHX3000 and KHX3200?

I'm thinking about geting an IC7 w/KHX3200.

[ThugsRook]

no i havent tested any KHX3200.

wouldnt it stand to reason that its the same stick as a 3500 tho?

[kevvv]

makes sense...:)

It feels strange to buy the old stuff thats rated lower and expect it to perform better...Do you reccomend the 3000 over the other hyperx?
Kev

[ThugsRook]

yes ~ for 533fsb cpu overclocking (dcddr) KHX3000 is the stuff to get.

Corsair XMS3000 was also "the good stuff", but i havent tried any in quite some time.

:-)

[kevvv]

ahhh....my system is a gonna be an 2.4C 800 FSB CPU...so does that change your opinion somewhat?

[ThugsRook]

well you saw the review ~ the KHX3000 can goto 425dcddr c222 max.

if thats in your OC range, then yea go ahead and buy a pair.

:wave:

[Chuck232]

Hey Thugs,

I've read all of both your P4P800 and this IC7 review/posts. And I've gotta say, it's the most comprehensive overview of either boards on the net!! So many combinations and stuff. I only wish you were testing a 200MHz FSB CPU. :yummy:

Anyways, actually I'm getting a new system soon and I'm stuck between the P4P800 Deluxe and the IC7. Price difference between the 2 is negligable after buying a separate NIC.

The only reason I'm hesitating about the IC7 is that over at the ABIT forums, people have been complaining about the onboard sound and the problems it has produced. It seems like the mic didn't work at all and the sound produced by the onboard sound was extremely staticy and contained loud popping noises.

So I was wondering if you or any of the IC7 users here have had this problem and if yes, did a sound card solve the problem?

Thanks a lot.

:)

(BTW, I hope this is ok for putting here. I mean a review should be more than just the performance right?;) )

[ThugsRook]

thx! :-)

both boards ran very nicely for me. i was using a SBLive5.1 on both boards. i dont use onboard sound or nic even if it comes with it.

i really liked both boards and neither caused me any probs. (any i couldnt fix anyways)

my choice for the IC7 over the P4P800 was purely a personal choice.
i had both boards sitting here and i had to choose one.
me and the IC7 just get along much better ;)

dispite the fact i gave both boards possitive reviews there are many ppl having serious problems getting their systems to overclock and run stable. even the guy i sold my P4P800 too is having problems.

ppl trying to duplicate my system arent having an easy time of it either.

both these boards are a little tricky when it comes to overclocking and best performance. a less experienced overclocker could easily have a hard time with it. (and even some more experienced ones :na: )

my results force me to recommend the IC7.

:wave:

<EDIT> P4 systems overclocked this high need extreme case cooling. 2 80mm case fans aint gonna cut it at all. 2 120mm case fans might ;)
(recently added another 120mm fan to my box)

[ThugsRook]

Thermaltake NB HSF...
http://www.thermaltake.com/images/products/chipset/productTiger1.jpg (http://www.thermaltake.com/products/chipset/tiger.htm)


i am completely surprised at just how well this HSF performs on this/these motherboards.

some of you remember me getting this HSF way back for my P4B533, and at the time it didnt work well at all.

i couldnt even try it on my P4G8X-D because the chipset was too big for this unit.

but on the i875s smaller core its working great and extreamly quiet!

:-)

[pastorjay]

I never really have used the heatsink on that thing. Used the fan on every board I have had in the last 8 months or so (which has been quite a few!). I am using the Zalman heatsink mod on my IC7-G now, and I love the low noise level on that! ;)

PJ

[ThugsRook]

yea i like that zalman mod ~ it looks sweet!

there are 2 reasons why i didnt go that route tho...
1) you cant put a fan on it if you wanted to (too small)
2) you have to reuse the stock "strap", which to me looks like its gonna rip the "U" prongs right out of the moboard :look:


im just very (surprisingly) happy with the TT-t1 and wanted to give it a shameless plug.

:-)

[eva2000]

interesting someone else i know, Creative over at xtreme and OCAU forums did the TTiger NB heatsink on his IS7 or IC7 and didn't make much a of a change in temps

[ThugsRook]

my NB temp only dropped 1*C ~ but i like the unit, its nice and quiet / easy to install / etc.

i didnt like the stock unit at all. i just plain wouldnt use it. (too loud) so i was using a big NB HS from a P4G8X with the TT 40mm fan.

i had to sell my P4G8X tho so i needed a new NB HSF, and TT filled the bill :-)

[Chuck232]

If you do the Zalman mod, the ABIT NB fan can still fit on top of that, probably creating much lower temps. Just use the fan speed controller to lower the fan speed giving you nice temps and low sound. :D

[ThugsRook]

appreciate it Chuck but i have no need. my NB temps are as low as they get and the TT fan is silent.

im a happy camper :-)

[Chuck232]

Ya, well, if you've got a TT t1, it's probably as good if not better than the Zalman and the stock NB fan...

Just an option for people who are gonna do the Zalman mod.

[eva2000]

i put 348cfm of ans over mosfet, NB, memory and video card and NB/system temps on PX865PE PRO2 dropped 10-20C from around 42-54C to 25-38C :D

[Chuck232]

:wack: :wack:

Lol.. must be pretty loud?

[ThugsRook]

LOL ~ yea thatll do it :lol:


my NB temps are 29-30*C max with my little tiny 5cfm 40mm 20db fan :yummy: hehe~

[eva2000]

kinda hehe http://www.fileshosts.com/pentium4/D1/albatron/px865pe_pro2_ich5/hsf/ich5_hsf_04.html

2x 92x38mm sunon 120cfm
1x 120x38mm panaflo 108cfm
1x 80x38mm delta 80cfm for cpu fan

[ThugsRook]

if your system ever catches on fire ~ the fans will put it out :rock:

[Mikki]

lol! :lol: :chit:

[Chuck232]

Dear lord...

Hope you don't go to LANs too much... People would probably just love that fan setup...;)

But it does the job!

[eva2000]

Originally posted by ThugsRook
if your system ever catches on fire ~ the fans will put it out :rock: lol .. the 2 92mm fans are connected to a 3.5" bay rheobus and the 80cfm delta is connected to a thermaltake fan speed controller so i can turn them right down to low as well :)

[SupermanCK]

so i guess your room gets pretty windy when you are playing games or running benchies huh??? lol..

[Mikki]

That's part of the gaming atmosphere....:D

[Chuck232]

It adds to the realism I'd say...:D

[ThugsRook]

Ram slots 1+3 vs 2+4 @ 184fsb
* P4 2.53B @ 3.5ghz 184fsb 368dcddr

> 3DMark2001 <
1+3 = 16518
2+4 = 16511


> UT2003 Demo < 1024x768, flyby --- botmatch
1+3 = 201.3 --- 92.4
2+4 = 201.3 --- 92.4


> Quake III v1.32 < 1152x864
1+3 = 320.4
2+4 = 320.3


> Aida32 <
1+3 = 5050 / 1771
2+4 = 5057 / 1771



not much difference between them.
slots 1+3 are more stable then 2+4 tho.

:wave:

[kevvv]

Thugs,

how are your temps on this board? forgive me if its in this massive thread somewhere.

Kevin

[Mikki]

Thugs, that's good news on the ram slot testing, was it the same on the Asus board? :)

[Hawk]

Interesting info on the RAM placement. . . .

[ThugsRook]

kevvv ~ temps are high on this board (in the 60S)
weather or not it is correct is debateable. ive run this same cpu on many different boards and they all read differently. ive learned to ignore it and not care.

IMHO: the temps are correct or at least closer then ppl think.

check this threads index (main post) for my P95 results. i always post board readings in that post in all my reviews.


Mikki ~ thats a big negatory!
only 2 ram slots worked stable on the Asus P4P800 ~ very strange :scratch: (check that threads index for the results)


Hawk ~ i learned long ago to always check each of the slots for stability. youd be surprised at what i find sometimes ;) (like on the p4p800)


:wave:

[ThugsRook]

CAS Tweaking
TEST SYSTEM
2.53b @ 3.45g 181fsb 362dcddr
Abit IC7 (canterwood)
Kingston HyperX 3000 2x512mb



> 3DMark2001 <

* NBs533 Auto

3.0-8-4-4 = 15925
2.5-8-4-4 = 15893
3.0-7-4-4 = 15927
3.0-8-3-4 = 16069 <
3.0-8-4-3 = 15976

2.5-7-3-3 = 16162
2.0-7-3-3 = 16169
2.5-6-3-3 = 16158
2.5-7-2-3 = 16306 <
2.5-7-3-2 = 16249

2-8-2-2 = 16320
2-7-2-2 = 16382
2-6-2-2 = 16388
2-5-2-2 = 16402 <



:wave:

[ThugsRook]

this is a strange one...

2.0-5-2-2 = 16404
2.5-5-2-2 = 16389
3.0-5-2-2 = 16410

:scratch: :scratch: :scratch:

[kONGO]

I would say that those are identical scores. There is an uncertainty with all benchmarks, they don't score exactly the same score every run. All those scores are within 1/10th of a percent from eachother.

There has been no testing trying to show how inaccurate benchmarks are. I did 15-20 (don't remember the exact number) of runs using SiSoft Sandra's unbuffered benchmark on my TUSL2-C with a 1.2GHz Cereal, and they showed a 3% span. This would mean that all scores within 3% can be considered statistically identical, since the tool (benchmark) is flawed (measurement noise is apparent in all measurements, benchmarks would probably not be considered very exact by a mathematician).
This span will probably decrease as runs are increased in number, but I suspect most benchmarks have an uncertainty of 1-3%. There are so many factors involved.

-kONGO

[ThugsRook]

oh yea? how about a retest then.....

2.0-5-2-2 = 16402
2.5-5-2-2 = 16414
3.0-5-2-2 = 16473

:scratch: :scratch: :scratch:

my benchmarks have never fluxed that much before!
notice the 2-5-2-2 settings is very consistant between the 2 runs?
thats were ive been running it. (and getting very consistant scores)

very strange.....

ill work on it more tomorrow :wave:

[kONGO]

I for one wouldn't mind a larger statistical study on benchmarks. I don't have enough mathematical knowledge in the statistics field, but that can probably be acquired with a trip to the library and maybe an e-mail or two (I have taken an introductory course in the stuff during my M.Sc. studies).

I've thought more and more about this the last few months, and have often found reviews who use a .5% advantage as conclusive evidence of performance increases to be slightly over the top.

I think that if you do 1000 runs with the same configuration with reboots every 5 runs you'll get a good picture of how the benchmark really fluctuates between every run. The code executed is the same, but the way the computer works isn't identical every run. Interrups and a things like that don't happen in a pre-defined way.

-kONGO

[ThugsRook]

kongo ~ i completely understand what youre saying, and im not even gonna debate it at all, BUT.....

cas3 faster then cas2? :wack: at 382dcddr? :wack:

no no no, somethings odd here ~ prolly an oddity in the GA Auto options somewhere when set to cas3xxx.

[kONGO]

Yeah, I was off on a little tangent, hehe. I tend to do that :D.

It does seem a bit odd, yes. My system's running at PC100 CAS2, so the concept of DCDDR382 is somewhat foreign to me ;).

-kONGO

[Mikki]

That's crazy....I've never seen them flux like that either...wth? :confused: Could it be the vdimm possibly? Or is it more than likely something the chipset is doing that the old code of 3DMark2001 can't handle properly? Odd.....:p

There's a lot of talk in different threads about benchmarks lately, we need to start a thread for that and get everyone involved testing this stuff...;)

[ThugsRook]

a quick recap of the CAS Tweaking post to show more of the same symptom....

3.0-8-4-4 = 15925 <--- faster :scratch:
2.5-8-4-4 = 15893

[Mikki]

A CAS problem? But with what, the memory controller or the memory itself? And what the heck is going on with it? Something's very loopy there...:rolleyes:

[kONGO]

How much has CAS had an impact on scores before? It may just be that at those frequencies, there are other factors that matter more than the CAS setting.

Since 3DMark isn't a memory benchmark per se, maybe it tops out before CAS has a possibility to show itself in the scores. That is, maybe the slight bandwith increase the lowered CAS setting can offer can't be utilized by the system when running the tests used in 3DMark and the differences in the score are due to 'benchmark noise'.

-kONGO

[eva2000]

Originally posted by ThugsRook
oh yea? how about a retest then.....

2.0-5-2-2 = 16402
2.5-5-2-2 = 16414
3.0-5-2-2 = 16473

:scratch: :scratch: :scratch:

my benchmarks have never fluxed that much before!
notice the 2-5-2-2 settings is very consistant between the 2 runs?
thats were ive been running it. (and getting very consistant scores)

very strange.....

ill work on it more tomorrow :wave: that is very interesting got my IC7 15 bios up with 2.4C and ocz pc3700 gold and my new 9600pros ..

[ThugsRook]

Mikki ~ the only thing i can think of off the top of my head is that the GA Auto (suboptions) are going wacko and everytime i reboot it gets a little confused and the Auto settings become slightly different then last time.

keeping the setting at c2522 does not cause a problem tho and its scores are very consitant.

ill work on it tomorrow :wave:

[ThugsRook]

eva ~ when you get the system settled in and ya have the time....

do me a favor run those same 3 tests please?

(c2 vs c2.5 vs c3) :wave:

[eva2000]

will do that.. will also be part of the big memory comparison shootout too :)

[Chuck232]

Hehe... well have you tried any other benches to see that perhaps it's just 3D Mark that's wacko?

[ThugsRook]

just to make sure my system is still stable i ran M86 for 10 hours last night (c2225 all 11 tests cache ON). it went 22+ passes and no errors.

im gonna work on the GA today to see if i can find something....

[eva2000]

try 5hrs looping memtest test #4 and 5hrs looping test #5 and then 4hrs looping goldmem full standard test should give you a better indication :)

[ThugsRook]

i think i have already found the problem....

Aida32

2.0-5-2-2 = 5058 / 1748
2.5-5-2-2 = 5056 / 1741
3.0-5-2-2 = 5053 / 1742

the 1st cas setting does not work?
maybe this is why you 250+fsb guys are having so much trouble on this board.

:na:

[eva2000]

oh my god so my pc3700 gold at 255mhz 2.5-4-4-7 is actually running stable at 2-4-4-7! :)

tried 1.4 bios ?

[ThugsRook]

i am not sure what its stuck @ to be honest with you.

i am looking into this further......

[eva2000]

well here's ocz pc3700 gold @233mhz 2-3-3-7 on Abit IC7

Sandra Max buffered memory (http://www.fileshosts.com/pentium4/D1/abit/ic7/memory/ocz/pc3700gold/233-233-2337-auto/mem-buff2.jpg) = 5536 / 5545
Sandra Max unbuffered memory (http://www.fileshosts.com/pentium4/D1/abit/ic7/memory/ocz/pc3700gold/233-233-2337-auto/mem-unbuff.jpg) = 3149 / 3186

and here's ocz pc3700 gold @233mhz 2-3-3-7 on Albatron PX865PE PRO2 with Enhanced memory performance (PAT) enabled using Sandra MAX3

Buffered (http://www.fileshosts.com/pentium4/D1/albatron/benchmarks/oczpc3700gold/2.8C/233-233-2337-hton-ehon-std/mem-buff.jpg) = 5386 / 5392
Unbuffered (http://www.fileshosts.com/pentium4/D1/albatron/benchmarks/oczpc3700gold/2.8C/233-233-2337-hton-ehon-std/mem-unbuff.jpg) = 3055 / 3142

and on the same Albatron PX865PE PRO2

2x 256MB GEIL PC3700 Golden Dragon @233mhz 2-3-3-7

Buffered (http://www.fileshosts.com/pentium4/D1/albatron/benchmarks2/2.8C/geil/pc3700gd/233-233-2337-ehon/mem-buff.jpg) = 5382 / 5382
Unbuffered (http://www.fileshosts.com/pentium4/D1/albatron/benchmarks2/2.8C/geil/pc3700gd/233-233-2337-ehon/mem-unbuff.jpg) = 2846 / 2859 <-- lower than OCZ PC3700 Gold!

2-3-3-7 is fairly close

[ThugsRook]

but are you sure the Alba was working right?

the p4p800 didnt work right at all using certain mem slots~
im rechecking mem slots right now on the Ic7.

[ThugsRook]

well, its not the slots.
both sets have the same symptoms ~ cas3522 is faster then cas2522.

still working on it....

[LastRide]

I believe it.GA is doing something weird with the ram (I guess cas settings) that it creats a problem.What if you flashed back to an older bios before they put in the GA option and test it?.

[ThugsRook]

here's something besides 3dmark to look @.....

UT2003 Demo flyby --- botmatch

2.0-5-2-2 = 200.7 --- 91.2
2.5-5-2-2 = 200.8 --- 91.3
3.0-5-2-2 = 200.8 --- 91.3

still working on it.....

is the Auto GA doing a "balancing act"?

[Mikki]

Looks like Thugs is at it again....:D

[Hawk]

Thrugsrook,

Interesting. . . . Maybe I can get some time next week to see that also on the ASUS board. I would expect to see a little more of a variance on the memory to score ratio. .

Hmmmmm. . . . .

[ThugsRook]

i dont know exactly what to make of this yet ~ but i do suggest running your own tests to make sure the cas settings you are running are indeed the fastest for all your benchmarks.

:-)

[pastorjay]

Man... I am gone for about a day and look what happens... Thugs has a major discovery... good work Thugs!

[ThugsRook]

take a closer look at the cas tweaking post and you will clearly see that faster cas does not always = faster system

the 1st score in each group is the base setting ~ scores below are 1 off, meaning only 1 change (in bold) has been made from the base setting.



3.0-8-4-4 = 15925
2.5-8-4-4 = 15893 <
3.0-7-4-4 = 15927
3.0-8-3-4 = 16069
3.0-8-4-3 = 15976

2.5-7-3-3 = 16162
2.0-7-3-3 = 16169
2.5-6-3-3 = 16158 <
2.5-7-2-3 = 16306
2.5-7-3-2 = 16249

[LastRide]

I ran 4 runs of 3DMark2001SE.Its intersting the more relaxed the cas the better the score but computer is less stable.At 3.0 cas the computer locked once and rebooted the other time both at 3.0 cas.GA is all on auto except for CPC is disabled.

2.8C @ 3.5GHz, IC7, 1:1 ratio DDR 500 250FSB

2.5/7/4/4 = 20,307
3.0/7/4/4/ = 20,340 ** 3DMark locked on first run.

2.5/7/4/3 = 20,381
3.0/7/4/3 = 20,536 ** Computer rebooted on first run.

[ThugsRook]

nice job ~ thx for the verification :wave:

[LastRide]

Tomorrow I'll run some cas 2 benches.I'll have to drop the FSB for 1:1 cas 2 and I will try 5:4 ratio aswell.I think we know whats happening already though.;)

[oldfart]

2-2-2-5
3Dmark 17987
Q3 402
UT2K3 237.94/92.1
TMPGEnc (lower is better) 4:23

2.5-2-2-5
3Dmark 17983
Q3 400.6
UT2K3 237.5/90.09
TMPGEnc (lower is better) 4:29

3-2-2-5
3Dmark 17816/17899/17876
1st run/2nd run/reboot-retest
Q3 400.4
UT2K3 237.54/91.45
TMPGEnc (lower is better) 4:28

3Dmark score @ cas 3 was all over the place.

CPUz 118c and ctiaw both report 3-2-2-5 as 2.5-2-2-5

[ThugsRook]

thx for the testing oldfart :-)
not as cut and dry as mine as LastRide's results, but still shows something is odd.

here's more of the same but this time i used the 4:5 @ 175fsb (438dcddr)

3DMark2001

2.0-8-4-4 = 15802
2.5-8-4-4 = 15878
3.0-8-4-4 = 15910



(hehe~ pc2700 @ 438dcddr) :rock:

[ThugsRook]

oldfart ~ i think if youll retest using X-8-4-4 your results will be much more obvious.

:wave:

[LastRide]

Here's a couple quick tests.No reboots or lockups this time.

2.8C @ 3.5GHz, IC7, 1:1 ratio DDR 500 250FSB


2.5/8/4/4 = 20,284
3.0/8/4/4 = 20,449 *

[ThugsRook]

nice :-)

im gonna continue to play around using the 4:5 so my memory will be up closer to the level you guys are running at. the difference is alot more obvious up here, might help me find something.

:wave:

[ThugsRook]

2 possibilities....

p1) the GA is doing a balancing act
setting cas 2.5 (or 3) is allowing the GA to apply extra options that it couldnt (or wouldnt) when in cas 2 (or 2.5).

p2) cas2 and cas3 are reversed in bios

[Chuck232]

:eek: Wow... this is something..... I will follow this closely! :eek:

[eva2000]

Originally posted by LastRide
I ran 4 runs of 3DMark2001SE.Its intersting the more relaxed the cas the better the score but computer is less stable.At 3.0 cas the computer locked once and rebooted the other time both at 3.0 cas.GA is all on auto except for CPC is disabled.

2.8C @ 3.5GHz, IC7, 1:1 ratio DDR 500 250FSB

2.5/7/4/4 = 20,307
3.0/7/4/4/ = 20,340 ** 3DMark locked on first run.

2.5/7/4/3 = 20,381
3.0/7/4/3 = 20,536 ** Computer rebooted on first run. very interesting indeed.. about to install rage3d for some 9600pro/9600pro vivo testing :)

[LastRide]

Here is a little more extensive testing.

233 FSB 1:1 3.26GHz 466 DDR

2.0/6/3/3 = 19,724
2.5/6/3/3 = 19,818 *
3.0/6/3/3 = 19,689

2.0/7/3/3 = 19,728
2.5/7/3/3 = 19,966 *
3.0/7/3/3 = 19,848

5:4 372 DDR

2.0/6/3/3 = 19,222
2.5/6/3/3 = 19,290 *
3.0/6/3/3 = 19,227

2.0/7/3/3 = 19,381 *
2.5/7/3/3 = 19,177
3.0/7/3/3 = 19,235

Cas 2.5 coupled with 7 is the fastest at 1:1.When I move to 5:4 things start to get a little sporadic.Cas 2.5 is fastest with 6,but with 7 its the slowest.I re-ran 2.5/7/3/3/ 5:4 again and the second run was 19,191 so its still slower.:wave:

[oldfart]

I did some more testing too. This is @ 257 FSB | 5:4 | GA settings auto

2-5-2-2
3Dmark 17987
Q3 402
UT2K3 237.94/92.1
TMPGEnc (lower is better) 4:23

2.5-5-2-2
3dmark 17983
Q3 400
UT2K3 237.50/90.09
TMPGEnc (lower is better) 4:29

3-5-2-2
3dmark 17816/17899/17876
1st run/2nd run/reboot - retest
Q3 400
UT2K3 237.54/91.45
TMPGEnc (lower is better) 4:28

2-7-3-3
3dmark 17858
Q3 392.8
UT2K3 235.4/89.95
TMPGEnc (lower is better) 4:31

2.5-7-3-3
3dmark 17760
Q3 391.2
UT2K3 235.22/89.80
TMPGEnc (lower is better) 4:32

3-7-3-3
3dmark 17756
Q3 391.8
UT2K3 235.33/89.54
TMPGEnc (lower is better) 4:32

2-8-4-4
3dmark 17414/17545
1st run/ reboot retest
Q3 385.9
UT2K3 233.16/85.48
TMPGEnc (lower is better) 4:34

2.5-8-4-4
3dmark 17569/17602
1st run / reboot retest
Q3 382.5
UT2K3 233/87.56
TMPGEnc (lower is better) 4:36

3-8-4-4
3dmark 17500
Q3 383.3
UT2K3 232.95/86.45
TMPGEnc (lower is better) 4:36


For the most part, no surprises. Faster timings = faster performance.

CPUz 118c and ctiaw both report cas 3 as cas 2.5 on any of these tests.

It seems when the settings are "unbalanced" like 3-5-2-2 or 2-8-4-4 some odd things happen.

3DMark is not a good bench to be testing this. The results vary too much from run to run. In a 17000 score bench, a 1/2% change is 85 points. That is well within (way lower) than a normal variance, The other benches are more consistant.

[Thor]

Originally posted by ThugsRook

p2) cas2 and cas3 are reversed in bios

:scratch:

Yikes, I'll test these out for ya as well on my IC7, but I'm going to have to wait for my OCZ 3700s to come in this week, as my main rig's memory was shipped out.