posted
OK, GEIL PC4400, 11/4/4/3, 2.7Vdimm, using a XP2500 0330 at 8X, I can run SuperPI all the way to 265FSB.... At 270FSB, SuperPI crashes with the "Not Exact In Rnd" error.....
At 3.0Vdimm, I got the exact same results, which may indicate the CPU or the Chipset is now the limit. I would guess it's the Chipset. I am using my Full Water cooled system, chipsets and CPU, running 25°C
Both sets of PC4400 exhibit the same results...
At 11/4/4/2.5, 3.0Vdimm, I can run SuperPI at 240, crash at 245...
We may need a Vdd mod...
-------------------- Too Many Computers,... Too Little Time ..... Com'on AM4! Posts: 27505 | From: Fire Island, NY | Registered: Feb 2003
posted
With my Geil PC4400 at 3,4,4,11, on a good day, I could get 214 FSB. Some days 210 was it . After doing the 3.0v VDimm mod, I could get 225 FSB.
Today, I tried decreasing the multi to 8x and running the memory at 120% just to test. Guess what, 214 is as high as I can get, even with 3,6,6,11 timings. I'm back where I started from before the 3.0v VDimm mod (CPU @ 8x178, mem 178x120%=213.6. I can boot at 10.5x230 synch, and complete Super Pi 1M, but at anything above 214 FSB synch, 3DMark2001 crashes back to the desktop. If I run asynch, I can go to 9.5x250 on the CPU, mem at 83% which is 208 FSB. Super Pi 1M completes and 3dMark2001 will complete multiple times without problems.
The mini test lead that I'm using for the 3.0v VDimm mod is 50.8 cm long and doesn't look to be very heavy gauge wire. Maybe I should shorten the test lead and substitute in some heavier wire? Maybe I'm still not getting the full 3 volts (high resistance? I'm guessing here). What do you think? I also need to test each memory module separately I guess.
Thinking of changing the name again to oldandfrustrated.
You have had the board to 250, so it isn't the board...
The GEIL is the Ultra Platinum correct...?
What I do is boot up at some safe multiplier and FSB. Then while running SuperPI, increase the FSB 5Mhz at a time using Clock-Gen... When it gets to the unstable point, SuperPI issues a error message... But I'm using W2K and XP may operate differently..... Strange, very strange indeed....
-------------------- Too Many Computers,... Too Little Time ..... Com'on AM4! Posts: 27505 | From: Fire Island, NY | Registered: Feb 2003
posted
Geil Ultra Platinum PC4400-2x256 meg dual channel kit, running W2K. I have no problem running Super Pi 1M at 10.5x230 synch and 9.5x250 asynch 83% mem. 3DMark2001, UT2003 benchmark, and games will crash back to the desktop if memory is ran above 214 FSB synch or asynch.
"I have no problem running Super Pi 1M at 10.5x230 synch..."
But then you say:
"3DMark2001, UT2003 benchmark, and games will crash back to the desktop if memory is ran above 214 FSB synch or asynch. 2001, UT2003 benchmark, and games will crash back to the desktop if memory is ran above 214 FSB synch or asynch."
OK, now I understand... SuperPI is not a valid test for stability for your system.....
Hmmm... I never tried anything else on these test systems.... Games are very tough.....
-------------------- Too Many Computers,... Too Little Time ..... Com'on AM4! Posts: 27505 | From: Fire Island, NY | Registered: Feb 2003
sorry to hear about your probs... I would check them with memtest 1 at a time at the highest obtainable FSB,but it sounds like you have a bad set of GeIL.I believe Stan had a bad set some weeks ago and had to RMA them...
I have 3 sets of the GeIL PC4400 and they will all run 11-4-4-3 and 2.7V @10x250FSB(played Doom3 for hours today)on my AN35N's....
Alan
-------------------- Posts: 1706 | From: Born in New Jersey | Registered: Feb 2003
posted
I got it whipped! I tested each stick of memory separately and could go 9.5x250 synch with no problem. Then, both sticks, single channel mode, 3,4,4,11 got 9.5x250 again with no problem. My Geil PC4400 will not run above 214 FSB in Dual Channel mode on the AN35. Single channel and it's off to the races.
Time to post another screen shot and end that other 7 page thread .
posted
You're confirming what I was already mulling around. Probably will see if I can deal with Geil directly to get replacements. After all, they do have a lifetime warranty.
quote: OK, GEIL PC4400, 11/4/4/3, 2.7Vdimm, using a XP2500 0330 at 8X, I can run SuperPI all the way to 265FSB.... At 270FSB, SuperPI crashes with the "Not Exact In Rnd" error.....
At 3.0Vdimm, I got the exact same results, which may indicate the CPU or the Chipset is now the limit. I would guess it's the Chipset. I am using my Full Water cooled system, chipsets and CPU, running 25°C
Bad news kids, I have now done the Vdd mod and got the exact same thing. Here's a rundown of the system:
AN35/N Vdimm and Vdd mods,Watercooled holding 30°C AQXEA 2500 0330 8X200 to start; CPU is good for 2.3-2.4Gig Vcore= 2.0V Vdimm= 3.0V Vdd= 1.92V GEIL PC4400 11/4/4/3
Pass SuperPI @ 8X265; crash SuperPI @ 8X270
It must be that CPU won't go above 270...
-------------------- Too Many Computers,... Too Little Time ..... Com'on AM4! Posts: 27505 | From: Fire Island, NY | Registered: Feb 2003
posted
After reading through the Geil forums about this dual channel problem, it seems "most" people are complaining about PC3200 not running right in dual channel or not running it's rating. I think I read that they've upped the CAS on newer versions anyway. Geil's fix is to raise the CAS Latency from 2.0 to 2.5 or 2.5 to 3.0, according to the situation. So, I guess I'm screwed since I'm already at 3.0 CAS.
Anyone know how to get CAS 3.5?
Also, if you look at the SPD ratings with CPU-Z Ver 1.24, it shows 2.5,6,6,11 @ 263 MHz and 3,6,6,11 @ 277 MHz. I guess that's what the ads mean that say "SPD optimized for Intel chipsets".
posted
Strange about the 3200 problems. Helped my sister-in-law build a system using a AN35ultra,an XP2800, and Geil ultra Platinum 3200. and with the exception of the Tras, it's set to spec. And that's at 2.7vdimm.
Speaking of spec, this Golden Dragon is spec'd as 2.5-4-4-7 at anything over 232fsb. And the spd reading is 2.5-5-5-9. With the 2.95volt vdimm mod, I have no problem running DDOL or SuperPi at 240fsb and timings of 2.5-4-4-11. One thing I find interesting though is, both cpuz 1.24 and Everest [Aida32's replacement] show a build date of week 6, 2002. And got it from the Egg back in March.
![[Big Grin]](biggrin.gif)
Com'on AM4!
![[huh]](graemlins/HUH.GIF)
. After doing the 3.0v VDimm mod, I could get 225 FSB.![[dono]](graemlins/fragend013.gif)
(CPU @ 8x178, mem 178x120%=213.6. I can boot at 10.5x230 synch, and complete Super Pi 1M, but at anything above 214 FSB synch, 3DMark2001 crashes back to the desktop. If I run asynch, I can go to 9.5x250 on the CPU, mem at 83% which is 208 FSB. Super Pi 1M completes and 3dMark2001 will complete multiple times without problems.
![[Banana]](graemlins/BANANA.GIF)
![[Smile]](smile.gif)
.![[Frown]](frown.gif)
![[Roll Eyes]](rolleyes.gif)
