Jump to content
Sign in to follow this  
cheezies

uwackme's bh-5 burn-in

Recommended Posts

yes of course, here you go:

 

* Timings: E-A-E-11-2-2-2-2-9-12-E-E-F-E(CPC)-2-3-3-3-3-3-4(alpha's)-AUTO-AUTO-AUTO-AUTO-AUTO-AUTO(slew rates)

 

* Volts: VDDR3,30 (overvolts itself like 0,02/0,04), VDD1,9, VAGP 1,8/80Mhz, Vcore 1,9, MP10.

 

* FSB250 (for now, just untill the errors with cpc on will go away) and then i'm going to up it in 5mhz increments. And then i'll repeat this whole procedure.

Share this post


Link to post
Share on other sites

Left it toasting at 250 today, be back to check on it after about 10 or 11 hours, hope it's ok.

 

Bah 120mm Fan sitting ontop of the sticks and NB it'll be fine. Makes a hell of a noise, case kinda hovers about an inch of the floor.

 

Anyway, off for a dirty weekend with the Girly this weekend so they're gonna get about a 48hr + memtest marrathon :D

Share this post


Link to post
Share on other sites

Nice results.

 

It is a CH5-BH6-BH5 for sure, I dont have any TCCD to test, but would be interesting to see if TCCD Samsung and BT-D43 Hynix would benefit as well.

 

Seems the effect is greatest on BH5/6-CH5 due to the design of the memory cells in the silicon. But there isnt alot of variation in DRAM design, so maybe the others can be improved as well.

 

Yes the idea of keeping the error rate low is that the program will take time away from hammering the ram to "tally" the error, and the idea is to hit the ram as hard as possible, so try to target the FSB at the point where errors start but not so high that ALOT of errors occur (as this would take away from the overall effect we are trying to acheive...HAMMER the SNOT out of the ram).

 

I can only hypothesize that this effort "scars" the memory cell capacitor in some way that improves performance at the memory-cell level. But it only goes so far, so permament failure level damage doesnt occur. I dont think below 3.6Vdimm and scorching heat we could ever HARM a BH5 (or BH6 CH5).

 

Simply having a 80-120mm fan hovering over the ram/nb/mosfet area blowing on it will keep the heat from ever reaching a dangerous level no matter how hard you push the settings.

 

There is, as in all things, a range of results. As Wizz has shown some BH5 is truly TRULY asskicking!!! So end results, the MAX, can range from 240Mhz @tight to 285Mhz @tight (LastViking and Macci have reached these levels). So even after all the burnin, we all get stuck with "Luck of the Draw" with our results.

 

 

I dont see why this wouldnt work on VIA, NF3 and NF4, its really just a matter of can the settings available HAMMER the SNOT out of the ram. The NF2 with Oskar's bios lets use on DFI slam just about everything to the @tight'est point possible.

 

Once you start peaking out in results, its time to venture into tweaking the drivingstrength/slewrate settings, etc which can be VERY time consuming but get a few more Mhz.

 

Latest experiments on the NF7 have shown CH5 256M sticks like DS or 1 SR of 2 better than the stock DS 3 SR 10, but 512M sticks of BH5 like DS 6 and SR 7 best. I still have alot of research to do on my own system, but there is gold out there still to be found in the settings.

 

Eventually Ill get a 939 and see what goes with A64, but Im totally ignorant about it at the moment. I want the dust to settle on NF4 before I go there.

 

I'd KILL for a PCI-e motherboard for Athlon, I really dont want to spend $500 on X800XT for AGP, but I also dont want to plunge into a still muddled A64 world. Plus, the damn 2600+ is an butt slapper....still.

 

Im really impressed with PCI-e though, it will be a serious change for the better. I think a Ahtlon PCI-e board would eliminate the whole concept of "3D stable" as all the timings issues of AGP-to-memory float away.

 

Note to DFI...maybe we can make an Infinity-e out of the current components to implement PCI-e for Athlon for short$. Gimme a call :nod:

Share this post


Link to post
Share on other sites

I wonder why my KHX3200 BH5s are rated at 2-3-2-x in default timings.... shouldn't they be 2-2-2-x?

 

SPD timings table in CPU-Z shows 2-2-2-x at 200mhz, but when i tried to run 2-2-2-11 on my system, it needs up to 2.9Vdimm to get the system stable. is this bios related problem or what? i'm using Hellfire's 3EG Rev 3 bios.

Share this post


Link to post
Share on other sites

@uwackme:

you said that oscar's bios gives us the real max results on our nf-2 board, but i'm wondering which bios this is (maybe you can post this bios here for us). Currently i'm using hellfire's great work with the 8karev2 bios, which got my settings that i posted above. So what bios is it, maybe it would give my nf2 just that little push that it needs.

Share this post


Link to post
Share on other sites

When I say "Oskar" I mean all the DFI bios's since he added the all the extra ram timings.

 

If you ever tried to tweak an NF7-S or a A7N8Xdel you'd know how limiting the "stock" NVidia bios's are.

 

I for instance cant BOOT an NF7-S at its maximum since I cant set the DS or SR settings until I boot windows and run WPCRSET. So I cannot run MEMTEST with various DS/SR settings to find the max since I have no way of changing them to test.

 

The DFI bios's... and Im on Hell's HKA rev2.... the last one he released, allow us to alter all these extra settings to find "The Max".

 

Only thing more I'd want is to break out the AGP-memory timings vs CPU-memory timings as seperate items, at the bios level, so we could really zero in on MAXIMUM CPU memory settings while leaving the AGP settings to what ever they can be...and the AGP settings are BELOW what the maximum CPU settings are, as illustrated by our ability to memtest/prime/SiSoft/etc bench at MUCH higher FSB's than we can do 3D stuff at.

 

I can memtest/SiSoft/2Dstuff at a good 10Mhz higher FSB than I can do 3Dstuff. The reason being the memory timings for the AGP and CPU accesses to memory are shared, but the Athlon is ALOT faster maximum than the AGP path is.

 

Remember..SPD is just parameters in the serial rom on the DDR stick that gives the "ballpark" rating specs of the stick to the bios. Each manufacturer puts in what they think will be best for the most customers....assuming NO KNOWLEDGE of anything, even the DELETE key to get into setup :nod:

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
Sign in to follow this  

×
×
  • Create New...