PC4000 Performance - Reviews vs Reality?

[H2OGun]

Just would like to share my experience on the above subject. I made a decision to upgrade from PC3700 Gold Quad to any good PC4000 modules after reading rave reviews and user feedbacks in the forums. And, I am sure there are others who are in similar position. Lesson learned: I would skip this and wait for the next one, probably PC4200. Why??

First, I bought Corsair TwinX-1024 PC4000 since it was available locally for USD395. After testing it intensively on both Asus P4C800E and Abit IC7-G, the max it will run is DDR508. Beyond this, MemTest and Prime95 will report errors very quickly (<5 mins). I tried all the known workaround: increase vCore/vDimm (reasonable limit), relax timing, etc. Nope, it just does not want to be pushed further. Even at the rated speed of DDR500, it was not error free.

Just in case you are wondering about my hardware, here are some facts:
- My PC3700 Gold Quad can run stable, error free at DDR490
- My P4 2.4c CPU can run 275FSB @5:4 with Corsair TwinX-1024 PC3500
- CPU , NB, GPU are all watercooled
- Good PSU - PC Power and Cooling TurboCool 510

Second, I contacted an online reseller who happened to have OCZ PC4000 Gold in stock for USD465. There are not many of them, are they? The guy was nice enough to perform some testing for me on the same motherboards and he later informed me that none of his modules can run beyond DDR518.

Of course, technically this is not a case for RMA. They can performed at the rated speed and for my case, it seems OCZ can perform better than Corsair. But for those looking for "beyond-the-spec" kind of performance because of all those rave reviews/feedbacks, you really have to consider the fact that a normal end user (like myself) may not get those "review-level" kind of performance?

[H2OGun]

Another update on the so called high end memory. This time I had a chance to test Geil PC4200 533MHz DDR Platinum Series Dual Channel Kit 2 x 256MB. The module is available locally in a very limited numbers. And, I managed to ask the shop for a test drive on my Asus P4C800E.

Result: the maximum it will run is 534MHz DDR (267MHz FSB @1:1). So, the good news is that it more or less did run at the rated speed. But, not a lot more!! Only 1MHz to be exact.

It seems to me that memory manufacturers have adopted speed binning to the MAXIMUM where the buffer is so thin in order to maximize their profits. As the results, we, the normal customers, will have to rely more on RMAs due to this very thin buffer.

Buyers Beware!!

[Steelwingz]

Guess I must have just gotten lucky, then...

I'm running my 2x512 Corsair TwinX PC4000 sticks at 260 fsb, 1:1 on my P4C800-E Deluxe completely stable; 24 hours of memtest86 passed, no errors, 24 hours of two instances of CPUBurnIn, no errors, 12 hours of prime95, no errors.

I've also seen posts on the abxzone of others similar successes with OCR PC4000 ram. Many reports of 260+ mem speeds. I think the best I've seen so far is a report of stable at 275.

[Hawk]

I got rid of the Corsair XMS 3700's the same day I received them. They were the worst set of memory that I have ever tried.

Now, yes the wait for 4200's can be good, it is not necessary because the 4000's definitly clock way higher then what that guy tested for you. The P4C800 w/ fully utilize those chips. Unfortunatly, the Abit board, cannot run faster then 255FSB at the moment w/out the help of SetFSB. (This should be fixed in the next BETA BIOS)

[Radelon]

My OCZ PC4000 EL 2x512 sticks are doin' DDR584 at 2.85v...... so max was 518 or something...... haha. And these aren't even the Gold, which is the higher rated stuff.

http://radelon.home.mchsi.com/el4000/292cpuz.gif

[madwelshboy]

I too am running my Geil PC4000's @ DDR520 without any problems whatsoever, have tried @DDR550 but had 1 or 2 probs (prob cos my board only gives me 2.8v for mem), hey Radelon, what cpu and agp volt are you using? And what board?

(Great oc BTW) :-)

[H2OGun]

Finally I received my OCZ PC4000 Gold yesterday and since then I have been busy testing it with the latest 875 motherboard, Abit IC7-G Max3. Guess what's the result is? Very disappointing. The maximum I can run with no errors on both MemTest86 and prime95 is at 253MHz FSB. Raising vDimm beyond 2.8v made matter worse since MemTest will report errors on Test #4 even at stock speed of 250MHz FSB. I just don't understand why they perform so bad???

[Tony]

I ran my 4000Gold yesterday on the max3 at 282fsb error free.The same ram hits 288fsb on the P4C800D error free.

Im testing 4 3700 Gold modules on the Max3 at the moment and they clock nowhere near as high and as tightly timed as when they are in the P4C or P4P.I feel the board may have some ram load issues maybe..to early to properly tell.

Also I have found upping the VDimm pretty much counter productive with this board.keep 2.8V the max and see how you go.

[Thor]

Hmm, I read on Abit-Forums, that you need at least 3.0vdimm to use 1:1 ratios on high FSB OCs on the MAX3.

[Tony]

Im not seeing that.
I am using 4 dimms at the moment and upping to 2.9V just creates errors.

I have a LOT more testing to do and will be able to tell better how it performs once im done.

[eva2000]

Originally posted by H2OGun
Finally I received my OCZ PC4000 Gold yesterday and since then I have been busy testing it with the latest 875 motherboard, Abit IC7-G Max3. Guess what's the result is? Very disappointing. The maximum I can run with no errors on both MemTest86 and prime95 is at 253MHz FSB. Raising vCore beyond 2.8v made matter worse since MemTest will report errors on Test #4 even at stock speed of 250MHz FSB. I just don't understand why they perform so bad??? chainbolt's Max3 review http://www.overclockers.com.au/article.php?id=205384 says that Spread spectrum is permanently enabled in the bios and can't be turned off... could be a factor in lower oc's for some people ? Macci at xtremesystem forum get's around 20-30mhz less of an oc'd with Max3

Tony ... have you had a chance to look into the Max3 bios to see if they can turn off Spread spectrum ?

[Hawk]

Funny how this was not mentioned in any reviews. :confused:


Keep up the great work tony.

[Thor]

I take back my last post. I guess people are having some issues with certain MAX3 boards and vdimm over 2.9. Seems like a vdimm fluctuation problem.

Abit Forum Link (http://forum.abit-usa.com/showthread.php?threadid=22312)

[Tony]

I will have a look at the bios Eva..thanks for pointing the SS issue out.

At the moment im NOT impressed by the max3, I have conducted over 14 hrs of benching and im nowhere near done.

Im sure my review will be a little different to the one on the [H]

[eva2000]

Originally posted by Tony
I will have a look at the bios Eva..thanks for pointing the SS issue out.

At the moment im NOT impressed by the max3, I have conducted over 14 hrs of benching and im nowhere near done.

Im sure my review will be a little different to the one on the [H] 14hrs of fun :D

there's been a huge price drop on IC7 Max3 in Oz, from AUD$400-412 to AUD$352-385

Spread spectrum has always lowered my max oc's on PX865PE PRO2!

[Thor]

Originally posted by Tony
Im sure my review will be a little different to the one on the [H]

That's what I'd like to see. I think the [H] review was nothing but crap usual - "it works out of the box" review. Thanks BT.

[Tony]

I have finished 3700Gold 4 module testing and im now doing the 3700 gold 2 module testing.

I can say i was a little disapointed with the max3 with 4 proven good modules.My humble P4P800D is a LOT better.

Anyway...14hrs + to go before i get to PC4000 testing..LOL

[Thor]

Egads! 14 hours of memory testing?!

BTW, I saw your post of the modded 1.2 MAX3 bios at ABXForums. Why doesn't Abit just include the damn setting in the first place?!

[Tony]

its looking like the testing is going to take a LOT longer.Im still on 3700 gold and i have 4000 and 3200 Plat to go :eek: :eek:

Regards the bios im unsure why they hid everything..but its there if you want it.

[eva2000]

Originally posted by Tony
its looking like the testing is going to take a LOT longer.Im still on 3700 gold and i have 4000 and 3200 Plat to go :eek: :eek:
welcome to my wonderful world right now :D

[Tony]

Difference is im actually doing a report on the board for OCZ but tagging the review with it.Thats why im doing so much testing.

[eva2000]

Originally posted by Tony
Difference is im actually doing a report on the board for OCZ but tagging the review with it.Thats why im doing so much testing. ah i see... hopefully your feedback to OCZ and subsequently to Abit makes future bios/boards from Abit that so much better :D

[Thor]

Hey BT, so what's the verdict? And any news on your modded bios to disable "spread spectrum" for the IC7?

I'm dying to try and use my OCZ PC4000 at 1:1 at 2.8vdimm.

TIA!

[Tony]

I hit this 265fsb wall many are hitting 1:1.It would boot at 285fsb 1:1 but only 265fsb was error free and stable.I also have 4 512dimms of 4000 now and can say the Max3 will run 4 512's at ddr500 250fsb 1:1 but im afraid not much more, 255 fsb was generating errors.All these modules will clock a lot higher so it is a board issue and not the ram.

Anyway, if 250fsb and 2 gigs of ram 1:1 is what you want then the max3 and OCZPC4000Gold should do it:)

Ps..the board doesn't like cas3 so use cas2.5

[Jim]

Originally posted by Tony

Anyway, if 250fsb and 2 gigs of ram 1:1 is what you want then the max3 and OCZPC4000Gold should do it:)

[/B]

Yikes, $800 in memory. :look:

[Thor]

Originally posted by Tony
I hit this 265fsb wall many are hitting 1:1.It would boot at 285fsb 1:1 but only 265fsb was error free and stable.I also have 4 512dimms of 4000 now and can say the Max3 will run 4 512's at ddr500 250fsb 1:1 but im afraid not much more, 255 fsb was generating errors.All these modules will clock a lot higher so it is a board issue and not the ram.

Anyway, if 250fsb and 2 gigs of ram 1:1 is what you want then the max3 and OCZPC4000Gold should do it:)

Ps..the board doesn't like cas3 so use cas2.5


I am going to confirm the same on my IC7, 265fsb is about the max I can do @ 1:1 memory ratio using bios 1.7.

And yes, the board definately doesn't work well with cas3, and even if I set it at that, my IC7 reports cas2.5 in cpu-z.