that config was fine but I was pushing beyond tmax I'm certain and the ambient temps have been taking it's toll ...especially folding all day everyday.
So ...I tweaked down a little. F@h takes a hit of a few seconds per frame but everything else is zippier now and benches run the same or close to what I was getting before. Temps are definitely down at least 1C so I will be making a few more tweaks to the hardware arrangement maybe this weekend.
This is most stable for me 24/7 with ambient temps starting to climb. I'm really tempted to reconfigure the pump right up to my waterblock to try and get more turbulance going inside. I'm supposed to be getting 2 new pumps from Dave (C-systems engineer) because of some crappy review samples that I got by mistake. He's been out of town so I'm still waiting on that.
I'm also considering removing the IHS and putting more lap ...and then recovering it with a fresh coat of AS5 or ceramique.
Load temps as of now are 41c/26c having fresh cleaned filters one week. I usually go two between cleanings.
Since updating to 0503 I notice that after about a day the cpu (htt) throttles back from 241 to around 226. The cpu frequency drops for no apparant reason but when I reboot everything is back to normal.
Got rid of the 0503 bios and downgraded back to 0402 ...things were stable again until the c-systems mag pump went out. My third one in a month.
I ordered a new D-5 as a replacement and I must say this is a far better pump ...I'll never recommend mag crap again.
Also, while waiting on the pump I decided to remove the IHS ..bad decision.
I'm bumping this thread cause I just ordered a brand spanking new opteron 170 that should be in around next Monday!!!
Thanks to Muddocktor for all of his patience and help!
I would hold on and cancel that order. Prices are about to be dropped considerably in July on the X2 cpu's which clocks just as good these days.
Yes I've been reading about that all day today. As it is I was ready to cancel the order when they had shipped it overnight. It's been sitting at fedex all day I just need to run by there in the morning to pick it up.
I hope I get good stepping after all this waiting ...sheesh.
Mac have you checked out the new rd580 from dfi? I will replace this mobo in the coming mos.
Oh yeah it's been out there a few weeks now. Reviews say it has more tweak functions than necessary ...if that's possible.
My opteron 170 came in and I've been working on it ever since. Mudd came over yesterday and we got his system together and I started working on my system again when he left. I didn't get a boot from it the whole day.
So today (fathers day) I tryed to boot and no go. Called Jim and woke him up (sorry for that Jim) and got busy on my system again. I had a fluid leak from the wc setup and it saturated the cpu receptacle. I switched over to mct-5 non-conductive fluid.
So I pull the motherboard and proceed in spraying it down with isopropyl alchohol and really spraying it well. I removed the switch card that is placed in the secondary crossfire slot and it was saturated with the fluid as well. I sprayed the cpu receptacle over and over until I was convinced I had cleaned the contacts.
Put everything back together and this is my first post. Been folding so far at 2680 with not a hiccup. I did try 2700 ealier and it ran fine in windows until I pressed restart then it wouldn't reboot. But I'm already higher than I was with the opti 175.
I'll post the stepping in detail later but I remember it being 0551 ...not the best but I'll be ok!
So Chris, your Mushkin couldn't run at 264 MHz? I did a little more playing with my 170 right before coming back to work and I got it up to 2650 MHz, apparently stable, but I took it back down to 2520 before coming back to the rig. I just didn't have enough time to verify stability before having to come back to the rig. I'm running a 166 divider and set it to 315 X 8. I also didn't have the time to take my vdimm mod off the board either, but I plan to remove it as soon as I get back in from the rig so I can try 2.6v on my Mushkin.
So Chris, your Mushkin couldn't run at 264 MHz? I did a little more playing with my 170 right before coming back to work and I got it up to 2650 MHz, apparently stable, but I took it back down to 2520 before coming back to the rig. I just didn't have enough time to verify stability before having to come back to the rig. I'm running a 166 divider and set it to 315 X 8. I also didn't have the time to take my vdimm mod off the board either, but I plan to remove it as soon as I get back in from the rig so I can try 2.6v on my Mushkin.
I think I can get it up to 263 @ 1:1 but that's as high as it goes. Maybe I need to drop the multi or something idunno. I did some toying around though and I get up to 240 | 2.5-3-2-8 @ 2.8v using the 183 divider so you may want to keep the vdimm mod afterall.
I can get mine to run nicely even at 2700 but it drops wu and runs them twice as long. I ran sp2004 and it wont pass not even at 2650 for more than 20 mins. And this board wont even allow 1t. It will pass the stress test for quite a bit but it crashes in IE and I loose wus.
You might try backing the htt back to 260-262 and running the ram full speed and see if you are stable then, Chris. I remember looking at my stepping on my 170 when I was installing the SI-120 and the Beast on it and I think I remember mine being an 0551 stepping too. I have a feeling that 2600-2650 will be around my max stable speed too. I just didn't have enough time to really do any testing before I came back offshore. The problem I'm running into with my Epox board is the lack of a mem divider in bios between 200 and 166. It's making it hard for me to max out my memory along with my processor speed. At least the old NF3 chipset doesn't puke at running 1T command rate at high htt speeds though.
LOL ...mine pukes alright!
I can run full speed up to around 258 like you said ...but that only gives me 2580 so I set the divider to get to 264. 265x10 craps core 0 in less than 10 minutes while core 1 keeps going strong even at 270x10. So I run the divider.
But if you want the 183 divider all you should have to do is run 200 at 9.5x ...you should be able to take that to around 295 right?
How much vcore can you give with that epox?
I'll test 265 with a few wu and see how we do. That rhymes ...lol
Alas, no half multi's in bios on my poor old Epox board. :bawling:
As far as vcore, I can set up to .2v over the stock vcore, which works out to around 1.52v with the processor loaded (.03v droop), so I can get just a hair more vcore than you with your board but not much more. With the old winnie proc I could get to 1.6v vcore but the Opti has a lower default vcore.
I'm up to 2690 right now ...I should be at 2700 tonight. I'm expecting it to be stable but not expecting much after that if any. 2750 would be nice however since it would take me up to 250 mem which would be awesome!
edit:\ well I was at 2680 rather than 2690. I went to 2690 and got too excited to get to 2700 a few hours later. After it heated up it shut down ...I mean shut down power down everything. After I cleared the bios I booted back up and worked my way up to 2600@1.375v + ov. It wasn't stable here before so I'll start burning and working my way up from here.
2690 was doing very nicely but if I can get the core temp down just a tad I'd be more comfortable at 2700 ...especially with the summer heat.
Ok ...I'm starting the burn-in again from 2600@1.375v which is a little lower than previous most stable. Seems to be cool and stable so far!
0
LeonardoWake up and smell the glaciersEagle River, AlaskaIcrontian
edited July 2006
Your comments about burning in and CPU vCore demands subsequently being lowered got me thinking. When I built computers 1, 3, and 4 in my signature, I spent a few weeks overclocking them and working up optimal case and CPU cooling. I never really went back to see if I could lower voltages - until tonight. Your burn-in comments made me curious. Heck, if any CPUs were burned in, it had to be mine. (3 X 820 with 900MHz overclocks each, heating my house whether I wanted it or not!)
Indeed, I was able to lower vCore in all three computers, not just by a little bit, but substantially. Starting vCore, all three D820 machines, was 1.425 (indicated in BIOS). They are all running perfectly now at 1.375 (BIOS), 1.33/1.34 measured by Asus Probe.
Your comments about burning in and CPU vCore demands subsequently being lowered got me thinking. When I built computers 1, 3, and 4 in my signature, I spent a few weeks overclocking them and working up optimal case and CPU cooling. I never really went back to see if I could lower voltages - until tonight. Your burn-in comments made me curious. Heck, if any CPUs were burned in, it had to be mine. (3 X 820 with 900MHz overclocks each, heating my house whether I wanted it or not!)
Indeed, I was able to lower vCore in all three computers, not just by a little bit, but substantially. Starting vCore, all three D820 machines, was 1.425 (indicated in BIOS). They are all running perfectly now at 1.375 (BIOS), 1.33/1.34 measured by Asus Probe.
Thanks!
Much obliged!
I wish I had this explained to me a long time ago it would have saved much grief. The ambient temps here are ridiculous compiled with our nutty humidity.
I discoverd an awesome burn in tool yesterday that I've been using. It looks and acts just like SP2004 (prime) except that you have the option to use gromacs to burn-in with. It's called ORTHOS Beta and it's been doing great. Especially since I'm near my limits ...I can run it (automatically on both cores) along with coretemp and monitor my temps. If temps stay fairly stable and far enough away from max t-case (49c on mine) then I substitute it with real gromacs. I haven't lost a wu since!
Congrats on your success Leo!
0
LeonardoWake up and smell the glaciersEagle River, AlaskaIcrontian
edited July 2006
Congrats on your success Leo!
Since my last post, I've lowered the comps' vCore settings down another notch to 1.3625. At 900MHz OCs each and at full load, the indicated vCores now are at or near defaults. When I initially worked up the overclocks, I had seen comments from overclockers stating they had equally high CPU clocks with default vCore. At the time, I just couldn't believe it. I still wouldn't believe if I didn't see it with my own eyes on three different machines right in front of me. I had always thought "burn-in" was just a myth. The room is now a bit cooler and the CPU core temperatures are now down by about 2*C. Wow!
Yeah I thought it was a myth too but since I've seen it at work I like it!
I can do 1.40v stable at 2650 now ...maybe better by next weekend. I find the longer I stress it the more stable it is when I move up the frequency.
I got up to 2700 but not stable ...well more stable than before but not gromacs stable.
So ...I'm starting over. It's probably too late to do a burn in at this point since the cpu is almost a month old but I'm giving it a shot anyway.
This is where I begin.
I'm setting my stock frequency (200htt x 10multi) to the lowest functioning voltage. No overvolting of the southbridge or the cpu. Ram is set 1:1 for now but ht is set to 4.
So as it stands now for the next 12 hours:
200x10 @ 1.150v 800ht x4
I'll bump the frequency until I can't go any higher every 12 hours after f@h burning and then once I can't go anymore I'll set it up to default voltage.
Now my question is ...does anyone have a clue as to the default voltage of an opteron 170? New has it listed as 1.30/1.35 ...and also a Denmark core ...wtf is that?
Up to 202 this evening. When I came home it was folding really cool at 100%. I think it still may work at this later date. I guess we'll know when the deal is done.
I think that when I go to default core I will set it at 1.30 since the board overvolts like a madhouse even when overvoltage is disabled.
Well I've made it all the way to 208 and it's Sunday morning. It feels like I still have headroom to keep on going. This requires sooo much patience and a week from today I am leaving town for a week. I'll just have to leave it folding while I'm gone and resume from wherever I might be at that time.
On another note ...I've been gaming every night at these low frequencies and it's not so bad at all ...the processor must not be doing a whole lot of work with the 7800gt.
Comments
So ...I tweaked down a little. F@h takes a hit of a few seconds per frame but everything else is zippier now and benches run the same or close to what I was getting before. Temps are definitely down at least 1C so I will be making a few more tweaks to the hardware arrangement maybe this weekend.
This is what I came up with:
memclock value B]200 mhz[/B
cas latency (cl) B]2.5[/B
tras B]8 clk[/B (auto)
trp B]2 clk[/B (auto)
trcd B]3clk[/B (auto)
trrd B]2 clk[/B (auto)
trc B]11 clk[/B (auto)
trfc B]17 clk[/B (auto)
trwt B]4 clk[/B (auto)
read preamble B]5.5 ns[/B (auto)
async latency B]7.0 ns[/B (auto)
cmd-addr timing mode B]1t[/B
cpu frequency B]238[/B
processor frequency multiplier B]x11[/B
processor voltage B]1.375v[/B
ddr voltage B]auto[/B
pci-express voltage [+1.20v]
vcore over voltage B]enable[/B
southbridge over-voltage B]enable[/B
peg link mode B]disabled[/B
peg buffer length B]longest[/B
ht link speed B]1000 mhz[/B
cas latency (cl) B]3.0[/B
tras B]8 clk[/B (auto)
trp B]2 clk[/B (auto)
trcd B]3clk[/B (auto)
trrd B]2 clk[/B (auto)
trc B]11 clk[/B (auto)
trfc B]17 clk[/B (auto) (17 in bios - 16 according to cpu-z)
trwt B]4 clk[/B (auto)
read preamble B]5.5 ns[/B (auto)
async latency B]7.0 ns[/B (auto)
cmd-addr timing mode B]1t[/B
cpu frequency B]240[/B
processor frequency multiplier B]x11[/B
processor voltage B]1.40v[/B
ddr voltage B]auto[/B
pci-express voltage B]+1.20v[/B
vcore over voltage B]enable[/B
southbridge over-voltage B]enable[/B
peg link mode B]disabled[/B
peg buffer length B]longest[/B
ht link speed B]1000 mhz[/B
This is most stable for me 24/7 with ambient temps starting to climb. I'm really tempted to reconfigure the pump right up to my waterblock to try and get more turbulance going inside. I'm supposed to be getting 2 new pumps from Dave (C-systems engineer) because of some crappy review samples that I got by mistake. He's been out of town so I'm still waiting on that.
I'm also considering removing the IHS and putting more lap ...and then recovering it with a fresh coat of AS5 or ceramique.
Load temps as of now are 41c/26c having fresh cleaned filters one week. I usually go two between cleanings.
bios: 0402
Marvel: yk51x86_v85153
Uli: IntegratedDriver2.20
CoolnQuiet: amdcpu_v1.2.2
Adi: none
This is not good.
I ordered a new D-5 as a replacement and I must say this is a far better pump ...I'll never recommend mag crap again.
Also, while waiting on the pump I decided to remove the IHS ..bad decision.
I'm bumping this thread cause I just ordered a brand spanking new opteron 170 that should be in around next Monday!!!
Thanks to Muddocktor for all of his patience and help!
I hope I get good stepping after all this waiting ...sheesh.
Mac have you checked out the new rd580 from dfi? I will replace this mobo in the coming mos.
My opteron 170 came in and I've been working on it ever since. Mudd came over yesterday and we got his system together and I started working on my system again when he left. I didn't get a boot from it the whole day.
So today (fathers day) I tryed to boot and no go. Called Jim and woke him up (sorry for that Jim) and got busy on my system again. I had a fluid leak from the wc setup and it saturated the cpu receptacle. I switched over to mct-5 non-conductive fluid.
So I pull the motherboard and proceed in spraying it down with isopropyl alchohol and really spraying it well. I removed the switch card that is placed in the secondary crossfire slot and it was saturated with the fluid as well. I sprayed the cpu receptacle over and over until I was convinced I had cleaned the contacts.
Put everything back together and this is my first post. Been folding so far at 2680 with not a hiccup. I did try 2700 ealier and it ran fine in windows until I pressed restart then it wouldn't reboot. But I'm already higher than I was with the opti 175.
I'll post the stepping in detail later but I remember it being 0551 ...not the best but I'll be ok!
I can get mine to run nicely even at 2700 but it drops wu and runs them twice as long. I ran sp2004 and it wont pass not even at 2650 for more than 20 mins. And this board wont even allow 1t. It will pass the stress test for quite a bit but it crashes in IE and I loose wus.
I can run full speed up to around 258 like you said ...but that only gives me 2580 so I set the divider to get to 264. 265x10 craps core 0 in less than 10 minutes while core 1 keeps going strong even at 270x10. So I run the divider.
But if you want the 183 divider all you should have to do is run 200 at 9.5x ...you should be able to take that to around 295 right?
How much vcore can you give with that epox?
I'll test 265 with a few wu and see how we do. That rhymes
As far as vcore, I can set up to .2v over the stock vcore, which works out to around 1.52v with the processor loaded (.03v droop), so I can get just a hair more vcore than you with your board but not much more. With the old winnie proc I could get to 1.6v vcore but the Opti has a lower default vcore.
Currently at 258x10@1.375 ...temps are aight.
edit:\ well I was at 2680 rather than 2690. I went to 2690 and got too excited to get to 2700 a few hours later. After it heated up it shut down ...I mean shut down power down everything. After I cleared the bios I booted back up and worked my way up to 2600@1.375v + ov. It wasn't stable here before so I'll start burning and working my way up from here.
2690 was doing very nicely but if I can get the core temp down just a tad I'd be more comfortable at 2700 ...especially with the summer heat.
Indeed, I was able to lower vCore in all three computers, not just by a little bit, but substantially. Starting vCore, all three D820 machines, was 1.425 (indicated in BIOS). They are all running perfectly now at 1.375 (BIOS), 1.33/1.34 measured by Asus Probe.
Thanks!
I wish I had this explained to me a long time ago it would have saved much grief. The ambient temps here are ridiculous compiled with our nutty humidity.
I discoverd an awesome burn in tool yesterday that I've been using. It looks and acts just like SP2004 (prime) except that you have the option to use gromacs to burn-in with. It's called ORTHOS Beta and it's been doing great. Especially since I'm near my limits ...I can run it (automatically on both cores) along with coretemp and monitor my temps. If temps stay fairly stable and far enough away from max t-case (49c on mine) then I substitute it with real gromacs. I haven't lost a wu since!
Congrats on your success Leo!
I can do 1.40v stable at 2650 now ...maybe better by next weekend. I find the longer I stress it the more stable it is when I move up the frequency.
Are you getting any more OC out of it yet Leo?
So ...I'm starting over. It's probably too late to do a burn in at this point since the cpu is almost a month old but I'm giving it a shot anyway.
This is where I begin.
I'm setting my stock frequency (200htt x 10multi) to the lowest functioning voltage. No overvolting of the southbridge or the cpu. Ram is set 1:1 for now but ht is set to 4.
So as it stands now for the next 12 hours:
200x10 @ 1.150v 800ht x4
I'll bump the frequency until I can't go any higher every 12 hours after f@h burning and then once I can't go anymore I'll set it up to default voltage.
Now my question is ...does anyone have a clue as to the default voltage of an opteron 170? New has it listed as 1.30/1.35 ...and also a Denmark core ...wtf is that?
I have a few days to figure this out.
I think that when I go to default core I will set it at 1.30 since the board overvolts like a madhouse even when overvoltage is disabled.
On another note ...I've been gaming every night at these low frequencies and it's not so bad at all ...the processor must not be doing a whole lot of work with the 7800gt.