Jump to content
Corsair Community

My ram does not run @ rated speeds/timings


Eroc

Recommended Posts

Help! I have been having a helluva time with my ram and board combo. My 2-PC3500C2 sticks will not run at rated speed on my Epox 8rda3+. I get MANY errors in memtest on test #5 when in dual channel at [B]200[/B] . I have to set it to 2.5-3-3-7. Here is what I have done so far: Epox 8rda3+: Tested (test #5) memory individually @ 200 2-3-3-7 (as rated): 0 errors Tested both sticks, single channel @ 200 2-3-3-7: 0 errors Tested Dual Channel @ 200 @ 2-3-3-7: 100's of errors Tested Dual Channel @ 166 @ 2-3-3-7: 0 errors Abit IC7G: Tested Dual Channel @200 and 217 @ 2-3-3-7 in my P4 machine: 0 errors Why won't the rated timings work in my Epox board UNDERclocked even?! I may need to bios update though I checked all of them and none of them addresed this sort of issue. All test were done at default voltages (the dimms showed up as 2.7V which I figured was more than adequate) with my Athlon 1700 running at 200*5 or 200*7 (to eliminate the processor as a variable). Any help would be appreciated.
Link to comment
Share on other sites

  • Corsair Employee
If the memory works in single channel but not dual channel, that is a board issue. You have proven that the memory works at its rated speed with another system. Leaving only the MB and CPU to question. Since you tested the CPU with the lowest mult to rule it out, that leaves the board or another hardware component in that machine
Link to comment
Share on other sites

[quote name='RAM GUY']If the memory works in single channel but not dual channel, that is a board issue. You have proven that the memory works at its rated speed with another system. Leaving only the MB and CPU to question. Since you tested the CPU with the lowest mult to rule it out, that leaves the board or another hardware component in that machine[/QUOTE] This is true...AND I installed some PC4000 into it and ran it at 200 with 2.5-3-3-7 in DC and it seemed to be fine. I believe a bios update and some Epox support is in order ;) Oh they are not Twinx sticks as I could not find any and just bought two singles. They are the same rev. and all the numbers match so I figure that is not the prob. Thanks for your help!
Link to comment
Share on other sites

  • 3 weeks later...
Well, I got my new Mobo in and I am in the same situation as before. Right now I am sucessfully running at 217MHz in single channel mode, but will not run Dual channel w/o errors in Memtest86. I even ran it on another AMD board...An Asus A7N8X and it was fine. One thing though...in both boards it did not read the SPD correctly. In my Epox board it read at turbo settings...2.5-2-2-7 and at optimized and turbo settings...2.5-3-3-8. Why was it not detecting at 2-3-3-7 at any setting w/o manual adjustments? Could this indicate a problem? Update: This is what Epox said: [B]"If a replacement board causes the memory test program to exhibit the same issue then the board is not the culprit. Yes, it could be a compatibility issue with the software or memory. RMA replacement will not solve compatibility issues. I do not recall seeing you mention any board problems (lock ups, stability). So I would not worry about the memory test results as you will just waste $ in shipping back and forth."[/B] What!!! The board is ok...must be ram...whatever! I tested Mushkin Enhanced PC3500 ram in it as well and it had errors too! I am not going to say "Oh well, it seems stable even though it errors in memtest86." What do you think RamGuy?
Link to comment
Share on other sites

  • Corsair Employee
Again, if the modules will pass one up at the tested settings it's not a module problem. And we only gaurentee them at the tested settings one up. In addition, The SPD will be set to the closest JEDEC defined values depending on how old they are that would be either Cass 2.5-3-3-7 at PC-2700 Spec or Cass 2.5-4-4-8 at DDR400 Spec.
Link to comment
Share on other sites

Well I talked directly to the support guy at Epox and he is going to try to set up the same test. He said he had not had probs. but he was testing hynix chips on a Buffalo PCB. I told him the prob I was having was with Winbond BH-5s on both the Corsair and the Mushkin. I will also check for a newer version of Memtest to see if software is an issue. Thanks RamGuy!
Link to comment
Share on other sites

  • Corsair Employee
It could be that you are just right on the edge of the limit of your CPU, after all I think you said it was a 1700+, and you are tryinng to run it way over spec. In addition, I would try them on another system and run [url]www.memtest86.com[/url] on them one at a time. But I do not believe that the problem is with the memory.
Link to comment
Share on other sites

Well in the test I run the CPU at stock...133*11=1463MHz and the ram asyncronous at 200MHz at 2-3-3-7 (as rated for 217MHz) and it fails. I am sure it is not the ram or the cpu. I am with Epox support to sort out a possible compatibility issue. I will keep you posted. Thanks again.
Link to comment
Share on other sites

  • Corsair Employee
This may be a chipset limitation. If the FSB of your CPU is 400 For Intel and 100 MHz for AMD, the memory speed supports only DDR 266. If the FSB of your CPU is 533 For Intel and 133/166 MHz for AMD, the memory speed supports DDR 266/333 If the FSB of your CPU is 800 MHz for Intel and 200 MHz for AMD, the memory speed supports DDR 333/400. Please try and set the system to these settings and test the modules one at time. CPU Freq: 133 Mhz Memory Freq: 125% or 4-5 Dim Voltage: 2.7 Volts Resulting memory Frequency DDR333 SDRAM Configuration: Manual SDRAM CAS Latency: 2T SDRAM RAS to CAS Delay (tRCD): 3T SDRAM Row Precharge (tRP): 3T SDRAM Active to Precharge Delay (tRAS): 7T
Link to comment
Share on other sites

I did the test you described and it passed. In single and dual channel. I even ran the fsb at 166 and mem. at 166 in DC however...with fsb at 166 and multi at 5.0x (830Mhz) and memory at 120% DC I got errors. In single channel. no errors. either stick. fsb: 166 multi: 5.0 CPU:830Mhz Mem: 200Mhz timings: SDRAM CAS Latency: 2T SDRAM RAS to CAS Delay (tRCD): 3T SDRAM Row Precharge (tRP): 3T SDRAM Active to Precharge Delay (tRAS): 7T Single channel: no errors Dual Channel: numerous errors (as much as 70 in one pass)
Link to comment
Share on other sites

  • Corsair Employee
Can you test the modules one at a time on another system? If you do find you have a bad module please follow the link in my signature and we will be happy to replace them. But from your previous post, and I: [QUOTE]Abit IC7G: Tested Dual Channel @200 and 217 @ 2-3-3-7 in my P4 machine: 0 errors[/QUOTE] That tells me the problem is not the memory. I would suspect that you’re maxing out the CPU or MB at these settings. And the fact they will run above spec in another system verifies that!
Link to comment
Share on other sites

I would have to agree. The memory is good. My friend's memory that is Mushkin Enhanced PC3500 (also Windbond BH-5s) has errors as well. AND...since my proc was underclocked for the tests (and does fine in single channel at higher speeds) it is safe to say it is not the proc. Also my board is an Epox 8RDA3+ Nforce2 Ultra400 and is rated for 400Mhz bus (200*2) so it SHOULD run at 200 if not more. I am working with Epox on possible compatibility issues with Winbond BH-5 chips and the 8RDA3+. Has anyone had this similar prob. or heard of one?
Link to comment
Share on other sites

[quote name='RAM GUY']No that would not be safe to assume! In fact, I would be suspicious of the CPU more so that then the MB at this point. You are trying to run a 1700+ at 3200+ speeds, is that correct?[/QUOTE] No. The multiplier is unlocked and in one test I ran the bus at 166 with the memory at 200 and the multiplier at 5.0. This makes the processor 830Mhz (166*5.0=830Mhz)...well under its rated speed. This proc has run fine at 2Ghz+ though in these tests I intentionally underclock the proc to be sure that it is NOT the culprit. So basically with the ram underclocked (200Mhz), the cpu underclocked (830Mhz), and the board memory speed set to its rated speed (200Mhz) at stock timings 2-3-3-7 and the boards bus below spec. (166Mhz): I get several errors in Memtest86 on test #5. My specs for test again: fsb: 166 multi: 5.0 CPU:830Mhz (Athlon 1700) Mem: 200Mhz (120% of fsb) in Dual Channel mode (PC3500C2) timings: SDRAM CAS Latency: 2T SDRAM RAS to CAS Delay (tRCD): 3T SDRAM Row Precharge (tRP): 3T SDRAM Active to Precharge Delay (tRAS): 7T
Link to comment
Share on other sites

  • Corsair Employee
You cannot look at it that way! The CPU was designed to run at 133 MHz not 200 MHz is the point. Sometimes you can get slower CPU's to run at a faster setting. However the only way to know for sure would be to a 2800+ or faster CPU and test the memory. In addition, with AMD platform you need to run the memory and CPU at the same frequency. If you run out of sync like that it will cause a bottle neck in the chipset and may generate errors that are not the fault of the memory!
Link to comment
Share on other sites

Ok...I can see what you mean. This chip has run at 200*10 with your pc3200LLPT mem. w/no probs...and has even run at 190fsb with you PC2700C2PT modules w/o probs. I have a Barton 2500 (multi locked however) that is designed to run at 166 that I may test. One other note. I do not get errors @ 166 or 200 if the module are in single channel; only in Dual channel. Syncronous, asyncronous, underclocked, etc. I agree that the memory should be fine. I will test the proc as that is the only thing that is still the same.
Link to comment
Share on other sites

UPDATE... Memory seems to run fine at 180 in sync with different proc. (testing Barton 2500) Still won't run async at 166fsb/200mem w/o errors. To compare I had tested this ram on an Asus board with a Barton 2500 with the same settings 166fsb/200mem with NO errors. So now I can probably say mobo though I am not done testing yet. There is a LAN party tonight and don't have more time to tweak around...just need a STABLE system.
Link to comment
Share on other sites

BUMP! I tested again: Barton 2500 @ stock (166*11) memory at 120% 200 timings 2-3-3-7. ERRORS in DC. Can't be memory as it tested fine in my Abit IC7G at 217 2-3-3-7 DC and in an Asus A7N8X fsb=166 mem=200 2-3-3-7 DC. Three different procs and three different boards (Intel and AMD) all with the same prob. only occurring in my Epox. I will contact Epox tomorrow AGAIN...
Link to comment
Share on other sites

  • Corsair Employee
Eroc, This is a limitation of the chipset. Let me put it a different way. With any AMD CPU at 166 MHz or above you would need to run the memory frequency at the same speed as the CPU. Or it will cause a bottleneck in the chipset and may generate errors that are not the fault of the memory. Intel P4 CPU has 4 single data rate data paths to and from the chipset to CPU. With dual channel Intel chipsets they will take the 2 of the 4 data paths and convert it to 1 double data rate data path per module. Thus dual channel DDR400 from and 800 FSB CPU! With AMD since is a double data rate data path to and from the CPU it’s more critical that the memory frequency be exactly the same as the CPU or 1-1.
Link to comment
Share on other sites

[quote name='RAM GUY']Eroc, This is a limitation of the chipset. Let me put it a different way. With any AMD CPU at 166 MHz or above you would need to run the memory frequency at the same speed as the CPU. Or it will cause a bottleneck in the chipset and may generate errors that are not the fault of the memory. Intel P4 CPU has 4 single data rate data paths to and from the chipset to CPU. With dual channel Intel chipsets they will take the 2 of the 4 data paths and convert it to 1 double data rate data path per module. Thus dual channel DDR400 from and 800 FSB CPU! With AMD since is a double data rate data path to and from the CPU it’s more critical that the memory frequency be exactly the same as the CPU or 1-1.[/QUOTE] Ok...then why does it work on an Asus A7N8X (Nforce) with 166fsb/200mem at 2-3-3-7 timings in dual channel with no errors then? It ALSO has a Barton 2500+ and has memory support up to DDR400 (Nforce) although its frontside max is 333Mhz. It is actually a slower board (333fsb vs 400fsb) than the Epox and it runs fine! If it is a limitation to the chipset it is particular to this Epox board then... I have done every thing I can to test not only between Intel and AMD, but between different mobos and processors. The only thing I think it could be is a compatibility prob. between the board and the Ram. Epox should be in the process of contacting you guys at Corsair to get some of this exact memory to test in one of their boards. I just hope you have some like it.
Link to comment
Share on other sites

  • Corsair Employee
I have answered this question! It may work and it may not. However I guarantee you if you run a benchmark with those settings you would get a reading of about PC-2100 for the memory even though you are running DDR400!
Link to comment
Share on other sites

[quote name='RAM GUY']I have answered this question! It may work and it may not. However I guarantee you if you run a benchmark with those settings you would get a reading of about PC-2100 for the memory even though you are running DDR400![/QUOTE] I'm thinking you mean PC2700 as the fsb is 333 (166). My Sisoft scores show that though they are understandably lower than when in sync. (Ex. 2522/2343 async.; 2565/2411 sync.) I do understand the importance of sync. operation in an AMD system. I tested the Barton 2500 syncronous at 200fsb/200mem and I got errors once again on test 5 of memtest86. I, along with you, believe the memory is fine. I also feel the cpus I have tested are fine. I feel the mobo has issues. Could you guys test an Epox board with my memory and see what you get? I am trying this with Epox and they are trying to get some Pc3500C2 Ram from you guys as we speak. Thanks for you help...I am not trying to be a nuisance I am just frustrated ;)
Link to comment
Share on other sites

  • Corsair Employee
Actually we have tested this MB a long time ago, and it was only rated at 333Mhz. However, you might go to [url]www.amdmb.com[/url], as they have an Epox section and I think there is some reworks that you can do to improve the voltage to the MB.
Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...