Asus A8V with OCZ3700 GOLD REV 3 - worth it?

[glorfy]

Test System as in my siggy.

I've been doing some test with my OCZ3700 Gold rev 3, with an Asus A8V Deluxe rev 1.02.
I was thinking of doing some benchmarks and typing them up. All my testing proved is that
it isn't worth doing. The memory seems fantastic - testing at 270FSB 3,4,4,7 - but this is
the problem - it would test stably at that, and then the next time it wouldn't. All it goes
to show is what we already know -the lock on rev 1.02 boards just isn't stable - although
I was fine up to 235FSB.

As I said, the memory itself was superb - even booting windows and benching at 2,2,2,5 -
although I would strongly advise against doing this. After doing this I raised to 205FSB!

LOL!!! I won't be making that mistake again!!! The system crashed - and the next thing I
see is a message saying "unable to find windows/system" - or some such! An hour later
after repairing my windows install, and enjoying the fun of digging out my USB FDD because
windows needs a driver disk for SATA drives, I found it was bed time and no more testing
for the day :-) I was ROFLMAO I can tell you - I consider it a lesson learned - Although I
didn't think that sort of corruption was possible!

Anyway, enough side tracking. I found my best/stable settings to be at default timings
(suprise huh?), FSB set to 233 with 10.5 multiplier, as I know 2.5Mhz is the most I'll get
out of my cpu.

As I've posted beofre I think OCZ3200 platinum would give me more with this motherboard,
due to lower latencies, but wasn't available. I think sooner rather than later my A8V rev
1.02 will be dumped. What to wait for though? I don't want to chuck away my 6800GT
when NForce 4, or via equivelent become available - but what's the point in upgrading to
a new motherboard, and not going PCI-e. It looks like I'll have to either sit and wait,
or spend more money than I really want to. I suppose secret hidden option number three is
to source some cheap intel stuff that's no longer current(478 2.4-2.8) stick the OCZ in
there, overclock thenuts off it, and leave it folding 24/7! I don't think the wife will
be in favour of that last option!

Conclusion.

The PCI/AGP lock issues with 939 motherboards have rather spoilt my enjoyment of my
system! It clearly stated on Via's web site that this chipset would have a working lock.
That being the case motherboard manufacturers having issues with it, is not acceptable.
I'm almost imploring with my motherboard to fall over so I can RMA it!

The OCZ ram is great stuff, and does exacty what it says on the tin, and what all the
reviews say it does, and as such comes highly recommended if you have a working PCI lock.
If you don't, stick some value stuff in there, as it will be more suited to the systems
performance.

[IceH]

Why don't you just try to rma it? Because they told you, it has locks when it doesn't? Isn't that some kind of deception?

[glorfy]

Sigh! I wish I could - according to via the chipset has a lock - Asus claim no such thing - there lays the nub of the problem. I should have done my research better. I f I'd had access to BE before I had my computer, I wouldn't have bought the board... A bit of a chicken and the egg situation - I could have done with net access first! Cheers for caring though! BTW I am still very happy with my purchases - just could be happier!

[BB_One]

Kind of where I am sitting with my new rig decision. 12 weeks of researches and forum reading an alas from the top 4 - 939 MB makers. None seem to deliver the kind of performance and aggravation free (to some extent), like the DFI UT 250 for example ~sadly the DFI is socket 754. :rock:

So I gave up on the present Mobo's and my new rig build is on hold. Positive impact however ...the new 0.90nm Athlon are now out so a 3500+ will suit me just fine thank u much. And hopefully the ASUS A8N-E Premium will be release in a month or so and be stable and fullfill all our dreams ( ya right). :bow:

So back to the drawing board and it looks like:

A8N-E Premium
Athlon 64 3500+ 0.90nm
OCZ 3200 Rev 2 2X512
BFG 6800 GT OC (oh oh not pci-e card ..rats back to drawing board) :beat:
Seagate Sata 120 gigs x 2
Audigy 2 ZS
OCZ Power Stream "620" if available by then if not 520.

:D :D :D


I wonder if I will see a major difference from my existing system consisting of a souped up PIII / Slot 1 with a tualatin 1.4 and 384 meg of memory and a ti 4200/128 meg at agp X 2.
:smash: :smash: :smash: nah, that sucker will make a lovely network router..LOL!

Ok.... are we in November yet? Can I order this stuff? Xmas is about to be here u know! :deviltail

[joeMan]

glorfy - sorry of your disappointment with the rev 1.02. I have the same board. I am sure I would be in your position if I were OC'ing my rig.

I have been reading about the ASUS A8V rev 2.0, and supposedly the PCI/AGP locks do work on them. Seems ASUS and/or VIA together should offer all owners of pre- rev. 2.0 boards an exchange if indeed they are guilty of having mfg. & distributing a product that was said to have been able to perform in a certain way and in fact does not...

I do have a question for you: In the BIOS of your A8V, are you trying to run those OC memory timings with a 1T command active? If so, this may be part of your problem. Try leaving it set at 2T. I have read where others have OC'd the hell out of them but only at the 2T setting. I only ask this because that setting in the A8V BIOS is kind of weird the way it's worded (enable/disable)...did you notice that...at the very bottom of the memory timings page?

[glorfy]

Ho Ho! No sign for a belly laugh maybe :)
<O>
Agreed on the memory I think the platinum is the choice for AMD - my etailer now has it in stock at around the same price as the gold rev 3 :smash:

How things move on! When I first started looking early July I was looking at an Amd 64 3200 with a 9800 pro - so at least I missed that mobo/Graphics card!

[glorfy]

Joeman -Nice to here from you again! Yah, all test were 1T, i.e disabled in the bios - EXCEPT that with this memory in auto 1T IS enabled. What is is with Asus - 1:1 becomes 2:1 :coocoo:

The thing is, sometimes the system would appear to run stably at say 270FSB. Another time it wouldn't run stable at 235! I can only put this down to the lock being fubar.

BTW - this is the link I was looking at
http://www.via.com.tw/en/products/chipsets/k8-series/k8t800pro/
To me, this is evidence that Asus have cooked up a substandard product. Whether it is the fault of Via or Asus, the product is not as described, and with this link in mind, if the board doesn't suppost a working lock - surely Asus should declare that openly

Any lawyers in the house?!
BTW that smiley in my last post was meant to appear as:
:)
<O>
Darn it just doesn't work!

[sierra_bound]

Asus never said there was a working lock on the Rev.1 boards. I don't see how that can be construed as deception. The A8V was one of the first 939 boards that came out, so people bought it, hoping the lock would work.

I bought my A8V when the board first came out. The VIA chipset does seem to have a problem with very high HTT (over 280). I've run 3700 Gold Rev.3 at 277 with 3-4-4-8 timings and no problems. I can run my 3700EB at 250X12 with 3-3-2-8 timings all day long. I have also used the Gigabyte 939 board. The VIA chipset is good for high CPU clocks, but the NF3 chipset is better for memory clocks. That has been my experience.

[joeMan]

Can you somehow force 2T and run the OC's again? 1T is awful tight for raising that HTT bus so high man and tweaking with the other timings as well.

Also are you giving that RAM enough juice?

-AND- is the CPU staying cool enough? ( or maybe the CPU needs more juice, or, maybe the CPU just can't scale!!)

-AND- are you sure your GPU can "HANDLE THE TRUTH?!" :lol:

-AND - one final -AND- ... since it's a 1.02, maybe you need to just find that ceiling, the rev 1.02 ceiling :( I know Iknow, but it's better than nothing :hide:
I'm talking about a nice, overall, kinder, gentler, George Bush Senior kind of overclock - where all the components work together for a succesfull "thousand points of light" overclock. The kind of overclock that's like where no one steps in dog **** and tracks it into the house at the birthday party kind of overclock. Maybe not as aggressive as you had wanted, but STABLE, and certainly everything performing HIGHER THAN SPEC.

[glorfy]

Guys! Don't get me wrong! I'm happy with my system, and I do like running a "George Bush Senior" O/Clock- maybe more of a "Tony Blair" overclock in my case though! Oops no - I'd probably be running at 4.77Mhz then, and get errors like"Cannot find file WMD". I thought it would be interesting to find the limits of the board as I "know" the memory will handle it. I've tried 250 with an 8.5 times multiplier without success - ie running at 2125. The fact is, sometimes it's stable with superpi and F@H and sometimes it just isn't - with the same FSB! I did read somewhere it's due to themethod they use for locking the clock - I just wish I could remember the link - if I find it I'll post it.

I only mention the RMA because IceH has. I bring up the Via link as proof that there should be a working lock, as the motherboard uses that chipset. I accept the fact that as far as high FSB goes - I've made the wrong choice - through ignorance - not through misrepresentation. I merely postulate that maybe somone should take responsibility for a screw up with these early boards.

I was probably pretty heavy handed in my first post - the idea wasn't to 'plead' for my board to fall over. It was to let you guys know how that particular board works with this pparticular memory in my experience.

Sierra-bound - thanks for sharing your experience with me - I don't imply deceit - merely that the chipset doesn't function as originally described by Via. If you purchased based on Vias description (which I did not) - then you would have every right to be peeved with the fact that you haven't got a working lock as described!

[glorfy]

PS - Joeman! Thats easy for you to say with your fancy, dandy FX53!!!Ooooooohh! :)

[joeMan]

Sierra Bound - Hurray for Sacramento! And, yes, nice advice for glorfy.

glorfy, I hope you get it ironed out man, i am pullin' for you. i hate it when i am workin on an issue and it keeps comin back in my face. none of my comments are meant to discourage.

Anyway - I must keep harping on the 2T issue, I really believe if you can some how force it and try again...

Also, which VIA Hyperions are you running?

[Deuce]

There is a new bios on the ftp. It's 1007 final. On the german ftp there is a beta as well 1008.001. The german ftp addy is ftp.asuscom.de
Try them out and see if it makes any difference.

[glorfy]

There is a new bios on the ftp. It's 1007 final. On the german ftp there is a beta as well 1008.001. The german ftp addy is ftp.asuscom.de
Try them out and see if it makes any difference.
Anyone tried these? As far as I know :) 1005.21 is the last revision to enable a lock on the 1.02 boards.... If anyone knows any different I'll try it! Cheers Deuce!

[glorfy]

Joeman - all advice is greatfully received! I have a rather 'dry' sense of humour - which most people mistake for 'no' sense of humour - I wouldn't be irked at you for trying to help - cheers! Anyway, to switch to 2T, go to the last line in the memory bios option and switch enable 2t to enable. I did some further testing just for you, and had some odd results. Anyway I had been using Clockgen for most of my overclocking, but decided to do it from the bios to eliminate any possible instabilities from there. Decide to start with default timings (2.5,3,3,8) at 10X240FSB. Booted into windows no problem. Ran Superpi to 1M in 35 seconds, no problem. Tried running superpi to 4M, faulted after 9seconds. Tried 2M same thing. Quit out of superpi, went back in, ran 2M in 81 seconds, then ran 4M in 178 seconds :( Whats that all about? Ran 3dmark03 and got by far my highest mark on test one - 338fps - then got a corrupted file fault. Tried to reinstall 3dmark03 - nearly finished and crash! Hmm.. not quite stable then - although this amy be related to 429/1130 o/c on Graphics Card. Anyway, while I was rebooting, I thought it would be a good time to try running 2T. Urrmmm - couldn't get into windows! tried three times, no joy. Went back into bios, changed to auto (1T with this memory 2T with corsair vs) and booted into windows. Ran superpi with no issues.

Several things come out of this.

1) Whats going on with superpi, and the strange behaviour?
2) I need to do more test with my GC clocks lower.
3) I need to test with Sierra-bounds' suggested timings.
4) I need to take it easy - I've got some folding to do, and I don't want to have to fix my windows install again!

[glorfy]

More testing! Lowered HT to 800. Set ram 2.5,3,3,8 with 1T. Went for 9.5x250. Seems stable. EXCEPT even though I've set 1T - I get 2T - not sure I understand this! Also SANDRA reports my settings as 264x9. Clockgen shows them as 250x9.5 - I'm confused! As it's now running 2T, its slower at superpi for instance, than it was at 235 1T. I suppose that's the price I'll have to pay, once I sort my cooling out to run a high FSB! What started out just as me posting how ram/mobo worked together seemed to be proving just how much I have left to learn!

Perhaps the lock is working stably - I just haven't learnt how to get the best out of it yet.

Thankyou all for your contributions so far! :thumb:

EDIT hit 255 with 2.5,3,3,8 that was max.
eased timings to 3,4,4,8 and hit 260x8.5. Tried to go for 266 and make 533 - that was a bit ambitious, and blue screened - with reference to usb.sys? Seems a bit strange - but the results are certainly better!

[sierra_bound]

One thing you should be aware of. When using a half multiplier, you are not gaining any bandwidth. This is because of the way the RAM dividers work. In order to calculate real RAM speed, you need to divide your CPU speed by the next highest whole multiplier.

For example, 260X8.5 = 2210 / 9 (next highest whole multi) = 245.5

[joeMan]

sierra bound, that's good to know about the half multipliers. Are you OCing your FX53? What BIOS are you using?

glorfy - glad you are making progress. nice you eliminated the 2T thing at the higher OC...although strange it is now popping up by default at slower HTT of 800 (hmmm ??). keep at it.

[Section8]

I have a revision 1.02 sitting in my closet as well. I couldn't RMA it because like you said Asus never claimed to have a AGP/PCI lock so I just bought the Rev 2 and I must say it is performing flawlessly!!!! It is the sh*t with this OCZ EL PC3200 Rev 2 Ram.

[BB_One]

It is the sh*t with this OCZ EL PC3200 Rev 2 Ram.

Ummm.....AMD + PC3200 EL rev 2. = above comments?

Perplexing and worrying comments that I have seen for the fourth time now, maybe not as crude but in same vein.

Anyone else agrees with that RAM to be more fun with Intel and frankly poor performer on AMD? Is this what some refer to as the BH-6?

[malinois1]

Ummm.....AMD + PC3200 EL rev 2. = above comments?

Perplexing and worrying comments that I have seen for the fourth time now, maybe not as crude but in same vein.

Anyone else agrees with that RAM to be more fun with Intel and frankly poor performer on AMD? Is this what some refer to as the BH-6?

No that ram is not BH-6. The OCZ 3200 Plat. Ed. Rev 2 is samsung TCCD chips. and with a 939 system will get around 290MHZ stable with 1t enabled ;)

[sierra_bound]

sierra bound, that's good to know about the half multipliers. Are you OCing your FX53? What BIOS are you using?
Running the FX @ 3GHz (250X12). Using beta BIOS 1005.021.

[BB_One]

Ha, thank you Malinois1 for the clarification...oufff... I thought I was back to the drawing board again with the ram.

Ouff what a relief....k now got to order 2 sets of Rev 2 before they do a dissapearing act from the market.

[glorfy]

Used the latest version of memtest86+(v1.27 final) for some further testing - http://www.memtest.org/-

It fails test 5 with 2 sticks at 233x9 -all other tests passed - I assume this is due to the memory controller. It passes both sticks single channel at that speed. Asus tell you to use the blue sockets - is there any harm in using the black sockets - as it passes 233x9 in dual channel when placed there. It will also pass 250x9, as long as you drop it to 2T.

I've read somewhere not to worry if an A64 fails test 5. Can anyone confirm this? If so why shouldn't I worry?

[Archangel]

One thing you should be aware of. When using a half multiplier, you are not gaining any bandwidth. This is because of the way the RAM dividers work. In order to calculate real RAM speed, you need to divide your CPU speed by the next highest whole multiplier.

For example, 260X8.5 = 2210 / 9 (next highest whole multi) = 245.5

Okay, based on that, can you explain something for me then? I can run and I'll put the pic in below, 10.5X250=2625MHz but IJ can't run 10X250=2500MHz. Also, when I try using official BIOS's, my video card give ut when I exceed 220HTT. So, why is it that my CPU will run 10.5X250=2625MHz but IJ won't run 10X250=2500MHz if the RAM divider rounds up to the whole? 10 is a whole the last time I looked. :irk: Also, where does the divider kick in? Both my RAM and CPU HTT/FSB show the exact same like they are in sync which is what CPU-Z and ClockGen tell me. Aren't they?

Curse my impatience with nForce 3. I think I should have waited even though I'd be corrupting my 3rd and 4th channel SATA drives OC'ing this much on one of them, but still. I would say my FSB/HTT are in sync. If htey aren't, please explain as to why not. Also, my locks don't lock when I use an official BIOS. :cry: Still, this 1005.021 does perform well. If anyone knows of a modded updated one, please post a link so I can give it a try. Please excuse the mess of my desktop in the pic below. :look:

http://www.lungehavencottages.ca/Desktop.jpg