*** Official ASUS P4C800/Deluxe (875P) Thread ***

[Xeon]

OK, guys, I'm going to ask one more time.

Are there any other benchmarking tools you would recommend I run and gather numbers for, other than Sandra and PCMark2002?

 

[jhites]

AIDA32
Spec ViewPerf
Seti Bench

 

[jhites]

Originally posted by: Xeon

jhites what makes you say that you don't put much faith in the Sandra benches anymore?

I've also noticed that most of you are referring to using Sandra 2003 and not 2004 (which is what I'm using right now). Is there something inherently wrong with the 2004 version of this tool?

Curious,

Back around April/May Thugsrook and many others of us did some pretty extensive benchmarking and testing here in a couple of threads. We basically found that the Sandra 2003 scores could not be trusted and had many variation curves. Using Sandra for cas testing is also useless. Not to mention the ODBC bugs.

Many of us ran various CAS tests using AIDA and were very happy with the results "curve", which seemed to be very accurate. Please feel free to test different CAS settings for yourself and compare. You might have also noticed that the Sandra benchmarks have been used less and less in the Reviews.

EDIT: On a Side Note - I have not updated or used Sandra 2004, so these issues may have been fixed.

 

[Xeon]

Originally posted by: jhites
AIDA32
Spec ViewPerf
Seti Bench

Where can I get Seti Bench jhites, it seems that there are a bunch of clubs but I see no place to download the tool.

I have all the other tools and then some.

Thanks!

 

[Xeon]

Originally posted by: jhites

Originally posted by: Xeon

jhites what makes you say that you don't put much faith in the Sandra benches anymore?

I've also noticed that most of you are referring to using Sandra 2003 and not 2004 (which is what I'm using right now). Is there something inherently wrong with the 2004 version of this tool?

Curious,

Back around April/May Thugsrook and many others of us did some pretty extensive benchmarking and testing here in a couple of threads. We basically found that the Sandra 2003 scores could not be trusted and had many variation curves. Using Sandra for cas testing is also useless. Not to mention the ODBC bugs.

Many of us ran various CAS tests using AIDA and were very happy with the results "curve", which seemed to be very accurate. Please feel free to test different CAS settings for yourself and compare. You might have also noticed that the Sandra benchmarks have been used less and less in the Reviews.

EDIT: On a Side Note - I have not updated or used Sandra 2004, so these issues may have been fixed.

Yes, as a matter of fact I started doing some digging with regards to this tool and found several other sites and reviews mentioning they couldn't trust the results so they weren't going to use it.

It's interesting that they would produce a tool that had these flaws to begin with.

 

[jhites]

Originally posted by: Xeon

Where can I get Seti Bench jhites, it seems that there are a bunch of clubs but I see no place to download the tool.
Thanks![/quote]There are several around with stats from many different systems to use for comparison.
You will need to install the CLC Seti Client / Driver
Then download the Benchmark File or this one here.
Both of these files have about the same AR of 0.416

My best completion time was 1.9525 hrs. @ 583.29 Mflops.

 

[CaptianAburrr]

I have been running between 488xxx to 497xxx on each client with the -o2 optimization with my main rig. My other P4 rig gets between 372xxx/385xxx on each client.

I added it up once today and I was getting like 980 total!!! at one point.. but on average 850

Run it from your floppy and let it do the standard auto start. The would be 7 of the 11 tests but will give a good indication for the memory stability.

I ran it again from the cd.. the normal auto standard tests and it returned like 53 errors..

Since you are running SoB, I would only consider the load vcore. I run my sig 2.6C rig with 1.5875vcore and it usually shows 1.488-1.520 with the fluxuation. This rig is a 2.6C @ 3.4Ghz (262fsb) with HyperX PC3500 DDR420 @ CL2-2-2-5 using 2.85vdimm. The mem is actually underclocked at but overspeced with the timings.

In bios i set my vcore to 1.525 and mem at 2.85v and it runs very stable now in windows.. however SoB still crashes a bit.. i think its b/c of the memory errors but i hate to clock down the speed and i hate to go any lower than 250fsb.. i actually got it to boot at 255fsb but not any higher and i messed with the voltages to get that.. it booted windows fine but memtest gave like 1000 errors on it..

I would not think so with your MCX and tornado. I am using a Panaflo 92mm 57cfm on the 900U and full load temps 46-48C running two instances of Prime95. Prime95 will heat is up more than running 2 instances of Seti or SoB. Two instances of these are about the same as your temps.

i read up some on the HS and thats what it seems they say.. they are nearly equal..

 

[computer]

John....

Normally, when you get the error windows will give you the option to send a report. One of the tabs will give you the location of the files. I know that does not give you the location but maybe the next time it happens you can trace it that way.

I got an INSTANT blue screen for about a tenth of a second then an INSTANT reboot. Error reporting never even saw it. This started after I removed the Raptor from the ICH5R connector and moved it to the Promise SATA connector. I DID NOT remove the drive from the device manager or clean the registry prior to that. I've been testing and swapping around various HD's without doing any cleaning or uninstall in the device manager, but that never seemed to cause any problem until I did the swap from ICH5R to Promise. I'm wondering if that could have done something it didn't like. This is just a test bed now, so I'm not concerned with the proper removal of anything or any cleaning since I'm going to reformat when I've determined which HD's are fastest and on what controller.

I'm going to post the results of all these benchmarks and tests, but I'll say for now I'm astounded and pi$$ed at one thing: EVERY SITE I could find, as well as posts at this entire forum say the ICH5R is faster than the Promise (or Si3112a) controllers even with a stand alone drive. I even saw HDtach results showing this. Well, it's not true. The Raptor on the alleged "dedicated speedy" ICH5R was only 77000kb on HDtach. I moved it to the alleged "PCI bus limited bottleneck chipset of the Promise controller" and got 85000kb!!! Furthermore, the Sandra HD benches are ASTRONOMICAL on the Promise controller!!! Over 1000mb/sec!!!! On the ICH5R, they are only about 95mb/sec!!! (This is the first number shown in the body of the tests, I forgot what it was since I lost the damn paper). Now, will someone please explain this??????? I wasted my $ getting this damn mobo version with the ICH5R controller, and it's SLOWER!!!!! Also, the Maxtor 80gb ATA100 7200rpm 8mb buffer and the Seagate of the same specs (except it's ATA100) BLEW AWAY the Raptor on the Promise controller!! Raptor was a bit over 1000mb/sec, Seagate was 1300 and the Maxtor was over 1600!! The Raptor did better though on HDtach on both controllers (Maxtor & Seagate respectively were about 62000k and 59000k on HDtach. Raptor was about 8.4ms average and other drives were about 14.1-14.2ms). Note though the Raptor is the OS HD and it's partitioned 4 times, and the Maxtor and Seagate drives are blank, so that's a disadvantage for the Raptor. (BTW, on my P4B533 and a Promise Ultra100TX controller, the results are OPPOSITE than that on the P4C800-E Deluxe!! The Maxtor beat the Seagate in EVERY test).

Another thing.....the Maxtor is ATA133. The Seagate is ATA100. The IDE connector on the Promise chip is ATA133. Logic would tell you that the Maxtor would be much faster. WRONG. In the tests where I dragged a 130mb file back and forth from the desktop to the HD on the Promise ATA133 connector, the Seagate was MUCH faster! WHY?? Yet the Seagate had lower scores in Sandra and HDtach! I'm bald over these benchmarks and tests.

Anyway, I have all the screenshots from the results and I'll post when I've completed them. I'm also going to add the WD800JB to the tests but can't now since it's what I'm using right now on this, my main PC.

SpecViewPerf....285mb download??? GEEEEEZE. Thank God I have a fast DSL line!

 

[computer]

Guys, on Memtest, how the hell are you supposed to make it scroll?? yeah, I know "CR", but what is that? Is that the "C" and "R" keys at the same time? That won't work, the config screen keeps coming up.

Why is taking so long to run? The Readme file says about about 20min. on a 366 Celeron, yet at 33min it's still running on mine! It keeps starting over, is that normal? It will got from test 1 to 7, the do it again, then again, etc.

The Readme doesn't explain what "Chan" means at the right side of the screen. Is that the memory channel?

EDIT UPDATE
Well, after 1 HOUR 15 MINUTES the SOB was STILL running! And that' SOB not SoB. The test just kept running and running. I noticed the error display kept moving; an error that would be displayed in a specific area on the readout on the screen, would disappear some other would replace it. Since I could not scroll to see them all, I don't know what was going on. Screen image (That's from a 15+ year old 13" CHEAP monitor I use for testing). Still getting all these errors at DEFAULT settings except for voltages. I even set PAT to auto and Turbo to auto or disabled if it was allowed. Something is ****ed somewhere. I ran memtest on my current PC (see the Asus in Rigs below) and after 15 min, not one single error, and it's overclocked! Both CPU AND memory! It also had better MB/sec readouts for the L1 & L2 cache that is shown on memtest!! RAM readout was a lot slower though on my current PC. I'm going to try the memory in my current PC in the 800-E Deluxe and see what happens.

If someone can tell me what "Chan" is I mentioned above, and if that is the memory slot, it appears what's in "slot 1" is bad.

 

[computer]

Has anyone ever tried the memory diagnostics tool from M$? http://oca.microsoft.com/en/windiag.asp

I changed the memory modules around (what was in the first bank I put in the second and what is the second I put in the first) and ran Memtest again. After 45 min. so far there is only one error this time! However, next to that one error the count so far is "8". What does that mean that there's only one error listed this time? Under "Chan" it still shows "1".

 

[Xeon]

Originally posted by: computer
Guys, on Memtest, how the hell are you supposed to make it scroll?? yeah, I know "CR", but what is that? Is that the "C" and "R" keys at the same time? That won't work, the config screen keeps coming up.

Why is taking so long to run? The Readme file says about about 20min. on a 366 Celeron, yet at 33min it's still running on mine! It keeps starting over, is that normal? It will got from test 1 to 7, the do it again, then again, etc.

The Readme doesn't explain what "Chan" means at the right side of the screen. Is that the memory channel?

EDIT UPDATE
Well, after 1 HOUR 15 MINUTES the SOB was STILL running! And that' SOB not SoB. The test just kept running and running. I noticed the error display kept moving; an error that would be displayed in a specific area on the readout on the screen, would disappear some other would replace it. Since I could not scroll to see them all, I don't know what was going on. Screen image (That's from a 15+ year old 13" CHEAP monitor I use for testing). Still getting all these errors at DEFAULT settings except for voltages. I even set PAT to auto and Turbo to auto or disabled if it was allowed. Something is ****ed somewhere. I ran memtest on my current PC (see the Asus in Rigs below) and after 15 min, not one single error, and it's overclocked! Both CPU AND memory! It also had better MB/sec readouts for the L1 & L2 cache that is shown on memtest!! RAM readout was a lot slower though on my current PC. I'm going to try the memory in my current PC in the 800-E Deluxe and see what happens.

If someone can tell me what "Chan" is I mentioned above, and if that is the memory slot, it appears what's in "slot 1" is bad.

You know, there seems to be a pattern here with MemTest86. I've seen several people from this thread that have complained about it not giving good results or just flat out wrong results.

I just tried using it and found that it's hanging on my box on particular tests and is taking way to long to run then it should. It's almost like it's incorrectly mapping the memory.

I've tried both the Version 2.9 and 3 with the same results.

I then tried Windows Memory Diagnostics and it runs without errors and gets through all the tests quickly. Now this doesn't mean that it's doing the same tests but it's starting to look like MemTest86 doesn't like this MOB and Memory combination that I have here as well.

I'm going to let MemTest run on mine over night and see how it goes.

computer CR means Carriage Return from the old days. Just an FYI.

UPDATE
I just had a thought, which could be part of the reason for the problems I'm seeing. At the Asus defaults, it's setting my memory timings to 3-4-4-8 but my memory defaults are really 2.5-4-4-7 so maybe setting these manually to the memory's defaults and any other settings that aren't really the hardwares true defaults will yield the correct results. I'm going to let mine run overnight with these set correctly and see how it goes.

Talk to you soon.

 

[Xeon]

Well it appears to be hopeless with MemTest on my sticks. It continues to hang on Test 6 no matter what I seem to set my BIOS settings too. Maybe there is some settings in the MemTest config that I need to set in order to get this to run right, I'm not sure.

At any rate I'd like to get to the bottom of this since if it's doing this due to bad sticks of RAM then I definately want to know about it ASAP.

It's really starting to bug me that this isn't working correctly and that there might be something wrong with these sticks, I find it hard to believe though because it's rock solid in normal use and since two other tools I have seem to show the memory is fine. I'm not sure what to think.

See ya.

 

[computer]

If that's what CR is, then I guess that's [enter]? Then what is "SP" the spacebar?

Well, I've come to one conclusion at least. My CURRENT computer is flawless. With it's current PC2700 memory it passed the Microsoft RAM test I posted the link to above with no errors at all, and it also runs Memtest with no errors at all. At least none after 36 minutes when I stopped it. I took the Geil PC4000 and put it in my current computer (P4B533) and it also ran Memtest AND the M$ RAM test with NO ERRORS AT ALL. So, the Geil PC4000 is evidently OK.....I guess it's safe to say that even though the memory is running slower on the P4B533??

Here's were it gets very confusing. My 800-E Deluxe cannot run Memtest with the Geil PC4000. It also cannot run the M$ test with the Geil. That M$ test is MUCH better than Memtest because it has loads of options plus it even tells you the bad memory bank. I thought I was on to something since it said the first mem slot/bank was the only slot getting the errors even when I switched the sticks around. I of course just knew then the mobo is screwed up. Well, I put only ONE stick in, and regardless of the banks used, it still gives errors with the MS test in the bank the stick is in!!!!! Any takers on that one? PLEASE?

There's more. I put my old PC2700 in the 800-E Deluxe and it ran Memtest PERFECTLY! Yet, it still fails the M$ test. Now if that setup would have also run the M$ test OK, this would be easy; it would be a memory compatibility problem. But, no, it can't be that easy. So, I can't figure out if I have a bad mobo, bad memory, or both.....or a bad CPU. I could not O'clock any higher with the PC2700 set to 266mhz. So I guess it's also safe to say my CPU is still crappy. ??

Michael, aren't you using the OCZ PC4000 Gold?
-Clint

 

[Rapamatic]

Ok, I got my board (P4C800-E Rev 1.02) good and stable with my 2.4 chip running at around 3.1, without all that much tweaking. But I have a few annoying things with my box that I'd like to get cleared up:

1) I put my CD and DVD burners as masters onthe two IDE channels on the motherboard, and would prefer to keep them alone on the channels. I have an 80 GB HD that I want to put up as a backup to my two striped SATA drives running off the ICH5R. So I plugged the one HDD into the blue promise controller, and in BIOS configured the promise controller to run as IDE, not RAID. At least upon bootup it seems to identify the drive, because I see it flash briefly on the screen. However, Windows XP does not detect the drive at all. Any ideas?

2) I bought a gig of the Corsair TWINX 4000PRO RAM (the one with the LEDs on top), but one stick doesn't light up at all (except for a few random flashes at boot, but nothing with activity, even under MemTest86), and on the other stick one of the LEDs seems burned out. The RAM itself is fine, 0 errors after 30 hours of memtest86. Is it worth RMAing to NewEgg? Is there some incompatability between my board and the RAM? I thought I read that the TWINX were tested on the Asus board, so I doubt it...

Thanks

-Ray

 

[jhites]

SpecViewPerf....285mb download??? GEEEEEZE. Thank God I have a fast DSL line

ROFLMAO - And Thank God their server is decent.

That being said: I will try to answer some of the MemTest questions. Since I run Seti and SoB, I want my system to be without error for the purpose of accuracy. I use ver3.0 on all my systems.

1. 1. Memtest86 is a very thorough testing program and will detect many errors that are often missed by other quick test programs. For the most part running the default 7 tests is good enough for most ppl. I use the default as I am overclocking to insure stability and then will run the extended when I think I have reached my max oc. The extended test which you use C-2-3-SP keys to get all 11 tests will take about 5-6 hrs to complete. By design Memtest86 will test as much memory as is possible. Even a CPU overclock that causes Windows to crash will most likely just cause Memtest86 to crash or hang, also. The hang or crash usually happens on the dreaded Test 5 or 6. This has all happened to me due to memory timings being to tight, fsb to high for cpu or mem, etc.

1. 2. MemTest86 repeats itself over and over until you esc the programs. It will continue thru as many passes as you determine is needed. Most say that 3-5 passes is sufficient. I use one pass as I move up the fsb or mem timings and then will use 3 passes as a general rule before going to all eleven tests.

1. 3. There have been many compatability issues with the i865/875 chipsets and memory modules. The Corsair LL modules come first to mind. There was even references to this on both the Asus and Corsair website at one time. On my P4C800-Dlx (not the E) I went thru several different modules before finally finding that the HyperX PC3500 which ran thru every test that I could run at 262fsb DDR420 CL2-2-2-5 (Performance Mode Enabled) with my 2.6C @ 3.4Ghz. I also used the TwinMos PC3700 on a P4P800-E with at 2.8C @ 3.3Ghz DDR492 CL2.5-3-3-7 (Performance Mode Disabled) that was also without error. I have Corsair PC3500 (non-LL Winbond BH5 modules) running a 2.4C @ 250fsb on a DFI PS83-BL @ DDR400 without error, as well. I of course bought most of my modules prior to the release of PC4000. The OCZ PC3700 Gold and the Mushkin Black Label were supposedly the best memory for the P4C800 from all reports that I have read.

1. 4. There have also been some reports that if you diable USB prior to running MemTest86 modules that were showing errors became errorless. I personally did not do this as I have 4 usb devices and want everything up and on when I run stability testing.

1. 5.

1. This is some notes which I found interesting at the mushkin site about testing that they performed on the ASUS P4C800 (not the E):
   
   

   The ASUS P4C800 is fully compatible with the Mushkin Level2 PC3200 and PC3500 DIMMS. The P4C800 will give the highest performance when the memory is running in synchronous mode. That is, especially with the new generation of 800 MHz FSB processors, it is advisable to only use the top grade of memory available.
   
   The P4C800 uses an AMI BIOS that does not correctly read the SPD and will, at DDR400 mode, automatically set the latencies to 2.5:4:4 (CAS: tRCD: tRP), at least with the current BIOS versions but an update may fix this. The ICH5 does not support RAID functionality, instead, an extra Promise PDC20378 SATA RAID controller is on board. Also, the 3Com Gigabit Ethernet controller does not support CSA technology.
   
   Important: If PC2700 memory is used in combination with an 800 MHz FSB CPU, it will run at 320 MHz instead of 333 MHz. The latencies can be set manually and we have been running the PC3500 without any problems at 2:2:2-6 in DDR400 mode with a burst length of 8 for highest bandwidth. At auto, the DDR voltage is already at 2.68V, the same was measured at the 2.55 and 2.65V setting. At the 2.75V setting, the board will supply 2.78V which is the highest we can tolerate without compromising the longevity of the DIMMs
   
   The P4C800 overclocks easily up to 300 (1200) MHz FSB, however, beyond 301 MHz, a dramatic performance hit occurs. Currently, the P4C800 is probably the fastest board on the market.

For your reference running my P4C800 using the 5:4 ratio with the (2X512Mb) PC3500 HyperX modules I get a mem bandwidth showing 2875MB/s in memtest. It would most likely be better if I could run 1:1 but no way at any timing these would do DDR524.

Computer: 27 errors on 11 passes is really not horrible, however 1879MB/s is rather low.

 

[gogoman]

Anybody elses (besides jhites ) with Kingston HyperX PC3500 (dual stick 1GB) able to run at 2225 or 2226 timings at DDR400 or greater? The best I can get is 24488 at 5:4 ratio at 250MHz(DDR400). Is all HyperX PC3500 the same?

 

[jhites]

Originally posted by: gogoman
Anybody elses (besides jhites ) with Kingston HyperX PC3500 (dual stick 1GB) able to run at 2225 or 2226 timings at DDR400 or greater? The best I can get is 24488 at 5:4 ratio at 250MHz(DDR400). Is all HyperX PC3500 the same?

--->Is all HyperX the same? <---Nope!
I am pretty sure that most of the Vendors are now using the newer Winbond CH5 chips which replaced the older BH5 chips. The Winbond BH5 modules (mine bought May'03 before CH5) were very good for tight timings but not higher fsb. The newer Winbond CH5 chips are much better for higher fsb but at more relaxed timings. My 2X256Mb of HyperX PC3500 bought at the same time will also do 2-2-2-5 with 2.85v on my P4C800 (<---I just sold this pair here on AT FS Forum ). My modules are KHX3500/512 2.5v with the blue heatspreaders.
My little self matched 1Gb pair

Thugsrook HyperX - Same stuff as mine only PC3000

 

[CaptianAburrr]

i downloaded a lot of bench and test apps today and im going to try them out tonight.. also.. any ideas on what i can do to get my computer to run stable at 2225 5:4 3ghz.. ? my ram is supposed to do 2225 at 434 i do believe.. it wont even boot at 1:1

 

[jhites]

Originally posted by: CaptianAburrr
i downloaded a lot of bench and test apps today and im going to try them out tonight.. also.. any ideas on what i can do to get my computer to run stable at 2225 5:4 3ghz.. ? my ram is supposed to do 2225 at 434 i do believe.. it wont even boot at 1:1

Hi Cap,
PC3200 does not spec at DDR434 2-2-2-5 and neither does PC3500 for that matter. There really is'nt any magic settings. It depends on your system and modules. I would start with 2-3-3-6 using 2.85v and just work my way down to 2-2-2-5. Change just one setting at a time. Like 2-3-2-6, then 2-3-2-5, 2-2-2-6, etc. I would also use infinite and 64uSec with all Performance modes set to Auto/Off. Put memtest on a floppy and leave it in drive A: to boot after each timing change. Run just one pass with each change. You will know when to quit by the error reportings. Wishing you luck.

 

[CaptianAburrr]

i just looked em up.. my ram (pc3200) specs for 400 @ 2335 and they also say they can run cas 2 at 434.. X-Bit got them to do 434 @ 2337 @ 2.6v.

i just ran pcmark 2002

scores:

cpu 7444
mem 9134
hdd 1925

are those decent?

 

[jhites]

Originally posted by: CaptianAburrr
i just looked em up.. my ram (pc3200) specs for 400 @ 2335 and they also say they can run cas 2 at 434.. X-Bit got them to do 434 @ 2337 @ 2.6v.
i just ran pcmark 2002
scores:
cpu 7444
mem 9134
hdd 1925
are those decent?

What speed and timings were you running?
I posted these test results a page or two back but I don't think they would be good comparisons.

CPU Score . . . . . 7882 . . . 8702
Memory Score . 10462 . . 10142
HDD Score . . . . . 1146 . . . 1191
The first set was running 3110Mhz @ DDR490 1:1 ratio and
the second set was running 3430Mhz @ DDR432 5:4 ratio.

I did run a bunch of tests some time back. Post your fsb, timings and speed and maybe I can look some up that would be better for comparison.

 

[CaptianAburrr]

I am running at 3.0Ghz with timings of 2-3-3-5-4 fsb = 250.. ram = 320/5:4

im going to reboot and take my cpu back to 2.4ghz and try and see how far i can get with 1:1 ram and 22254 (if possible..) and then see if i can make it somewhere where my bench is higher and i'll probably stay there..

 

[Xeon]

Originally posted by: jhites

That being said: I will try to answer some of the MemTest questions. Since I run Seti and SoB, I want my system to be without error for the purpose of accuracy. I use ver3.0 on all my systems.

1. 1. The hang or crash usually happens on the dreaded Test 5 or 6. This has all happened to me due to memory timings being to tight, fsb to high for cpu or mem, etc.

Well this isn't the case for me, I've set my BIOS to both the Asus defaults and to the defaults of the hardware and this still does not work, so this isn't the reason. I have a hunch I know what it is and I'll touch on it below.

Having said that, how in the world should I loosen things up in order to get MemTest86 to work with this setup if I've already done the test run with the defaults?

1. 3. There have been many compatability issues with the i865/875 chipsets and memory modules. The Corsair LL modules come first to mind. There was even references to this on both the Asus and Corsair website at one time. On my P4C800-Dlx (not the E) I went thru several different modules before finally finding that the HyperX PC3500 which ran thru every test that I could run at 262fsb DDR420 CL2-2-2-5 (Performance Mode Enabled) with my 2.6C @ 3.4Ghz. I also used the TwinMos PC3700 on a P4P800-E with at 2.8C @ 3.3Ghz DDR492 CL2.5-3-3-7 (Performance Mode Disabled) that was also without error. I have Corsair PC3500 (non-LL Winbond BH5 modules) running a 2.4C @ 250fsb on a DFI PS83-BL @ DDR400 without error, as well. I of course bought most of my modules prior to the release of PC4000. The OCZ PC3700 Gold and the Mushkin Black Label were supposedly the best memory for the P4C800 from all reports that I have read.

In all the cases you mentioned above they are all NON PC4000 sticks. I have a sneaking suspicion that MemTest86 isn't mapping the PC4000 sticks properly and is the underlying cause of people who have tried using MemTest86 with these sticks indicate that they have problems. The OCZ PC3700 and the Mushkin sticks are some of the better sticks out there for this board but the OCZ PC4000 Gold has been proven to be better than either of these two. It's unfortunate that during the last "Searching For The Memory Holy Grail" done here at AnandTech they didn't cover the OCZ PC4000 Gold sticks because it would have shown the results and in fact the OCZ PC4000 Copper was the one that proved to be the fastest out there, beating the 3700 Gold.

The PC4000 Gold in many other reviews I've seen has been the best in every board that's been used (See the other review below which isn't an Asus Board). I suspect now that the PC4200 is out, that it too will prove to be the best out there. See the reviews below for more info (I can provide a few more if you'd like so just let me know):

OCZ PC4000 Gold Review 1
OCZ PC4000 Copper Review 2

I'm going to be contacting both the developers for MemTest86, Asus and OCZ to let them know the results I am getting with this setup and have them look into the issue further. Even if it takes pulling my two sticks of OCZ PC4000 and sending them to OCZ so they can try these same tests on an Asus board running MemTest86, that's what I'll do in order to get to the bottom of which of these things is causing the problem.

I have tried several other Memory tools to test these sticks and they seem to be fine. I do however have concerns that these other tests may not be covering the way that MemTest does so further investigation is in order, until the truth comes out.

Thanks for the Info jhites, I really appreciate it.

 

[jhites]

Originally posted by: CaptianAburrr
I am running at 3.0Ghz with timings of 2-3-3-5-4 fsb = 250.. ram = 320/5:4
im going to reboot and take my cpu back to 2.4ghz and try and see how far i can get with 1:1 ram and 22254 (if possible..) and then see if i can make it somewhere where my bench is higher and i'll probably stay there..

The closest test that I have is 250fsb 5:4 DDR400 with 2-2-2-6 timing. Remember that I am running a 2.6C so there is a difference of +250Mhz in cpu speed which will also have an impact on the memory. With that in mind it looks like your benchmarks are looking pretty good for the settings. Your Raptor Raid blows my old Maxtor ATA133 IDE away.
8066 . . . . . . 9592(5:4) . . . 1206

 

[Xeon]

Originally posted by: CaptianAburrr
i just looked em up.. my ram (pc3200) specs for 400 @ 2335 and they also say they can run cas 2 at 434.. X-Bit got them to do 434 @ 2337 @ 2.6v.

i just ran pcmark 2002

scores:

cpu 7444
mem 9134
hdd 1925

are those decent?

Here are some I just did a day or so ago. Not sure how much it will help you though because these are all subjective based off the machine you run them on and the equipment you have.

CPU Score - 7922
Memory Score - 10585
HDD Score - 1360
Running 3239 Mhz @ DDR537 at a 1:1 Ratio


UPDATE
Equipment:

Asus P4C800-E Deluxe REV2.00
Intel P4 2.4C
OCZ EL DDR PC-4000 Dual Channel Gold/500 Mhz/2.5-4-4-7

Happy O'Clocking!