so you leave it alone with affinity on both cores eh?
This is as hard as I can seem to push my cpu on air.... It's only 51C. I'm at 1.425 vcore + 104% special vcore. If I go to 2.6ghz it's highly unstable. I'm not sure what preventing me from going higher.
What's the highest voltage I can safely run for operton 170's? I thought 1.45v is about the highest.
Have you tried to reduce your HTT multiplier? (sometimes called LDT multiplier)
That can create a hard wall unless it is reduced at around 240MHz reference clock. Since you are at 250, I'd recommend the 4x HTT multiplier (sometimes listed at 1600/800MHz HTT in the bios)
Definitely. I don't think I would even POST if I had normal. I'm using a x3 LDT multiplier. I increased voltage to 1.4 + 110% @ 261.. It's almost stable. 1.52v seems scary? Am I going to high with the vcore?
I can't seem to get past this to even boot up. I've tried 300ref x10 @ 133mhz=200mhz dram but it won't even boot windows. Strange thing is cpu-z won't even show the DRAM speed.
I'm using 1.2v+136%special for 1.632vcore. I increased ldt to 1.3v. I can't even pass Prime at 2.9 with these settings. The revision on top of the IHS is CCBWE.
The vapochill says it's at -48C. Any ideas?
The memory frequency is not being displayed because of the 143MHz divider. CPU-Z does not recognize that one for some reason. It does the same thing on my system if I try that one.
Did you get more info from the CPU stepping? Is it a CCBWE 0551 or 0550 etc?
In my experience, phase-change will buy you about 300-400MHz over air-cooling. Since you were stable at 2500MHz, I'd say a prime-stable 2800-2900MHz sounds about right for that chip. Could be that you are simply reaching the limits of the chip. My Opteron 148 did about 2.85GHz 100% stable on air, and about 3.2GHz totally stable on phase. You are lucky that you didn't get stopped short by the cold-bug too! Quite a few of these chips don't behave well at sub zero temps.
Could you take a quick A64Tweaker 0.6x screenshot for me? Perhaps we can find something holding you back there..
Do you you know the max safe voltage I can go? 1.63vcore seems really high and dangerous.
It's CCBWE 0543TPMW.
Actually, 1.63 is not that bad for phase. The maximum 24/7 voltage I'd be comfortable with on a 90nm AMD is about 1.65V actual. I've seen people do 1.65V on air/water too, and I know people have run chips for years without issues at that voltage. Even the vapoLS will start to choke a bit at higher than that. You can go a little higher, but I wouldn't recommend doing it for very long.
I'll see if I can find some info on that stepping..
So 1.7 even on phase would be dangerous? How much voltage would kill it? 1.7,1.8?
I'm currently at 1.5vcore+110%special=1.65vcore. Vapo still says -47C. What's strange is prime runs for about 5mins on the 2nd core, but it always fails within 3 seconds on the 1st core.
I locked in memory timings @ 3448 in bios.
So 1.7 even on phase would be dangerous? How much voltage would kill it? 1.7,1.8?
That all depends I guess. I can't say I've ever tried more than 1.7, but in my situation, the temps increased so much from 1.6 to 1.7 that it was pointless and actually worsened stability. Another thing to keep in mind is vcore stability. If your vcore jumps around at all, the danger is much greater. If you set 1.75 and it jumps around between 1.75 and 1.85, you could do some damage on one of those high spikes. I'm willing to venture a guess that you will not gain anything in stability by increasing above 1.65V..
It's generally accepted that 1.65V is about the 'safe max' so long as temps are kept in check. Any more than that is a risk, regardless of temperature.
About your stepping: I saw several examples of CCBWE 0543TPMW doing 2.6GHz with about 1.5V on air-cooling, so perhaps there is some more tweaking you can do to improve stability. Memory timings can make a world of difference (even when on a divider). Can you post an A64 tweaker screenshot?
OK, cool.. couple things to try, but first, set 133MHz divider in the BIOS (only temporary for testing, we'll increase it later)..
1) Set trc = 9 and trfc = 16
2) Set twtr = 2, retry to see if Prime stable..
3) Set DRAM Drive strength = 7 and DRAM Data Drive = 2 and try Prime again
4) Set DRAM Response Time = Normal and try again.. Also try 'Fast' but avoid 'Fastest'.
5) Try to increase Max Async Latency = 9ns or 10ns
6) Try different Read Preamble settings, try to increase/decrease a notch or two.
OK i'll go try that out right now. Before I do though, I noticed when I set FSB to 300x10 @ 133divider at 1.65 & 1.6775vcore it will list PCI devices but freeze after that. It won't show the "verifying dmi pool" and then "backing up cmos....complete!" It just hangs after pci device listing.
I also remember reading on dfi-street on a tutorial with setting up the bios, they said to disable bank interleave. What's that for?
edit: When setting it to 133 it freezes after pci device listing as well. I have to use 140 to get past the bios to boot windows.
Bank interleaving improves DRAM performance quite substantially. I haven't run into a situation where disabling it helped anything, but you could give it a try. It will hurt your performance though.
In regards to the early freezing in the BIOS at PCI device listing, I've seen that happen when the HTT is too high/low, or if the system is very unstable. Are you currently at the 3x HTT multi? Although it is not recommended, I sometimes find that using a higher than usual HTT works well on phase-change. 11x291 for me worked best with the 4x LDT multiplier for some reason. Wasn't the case on air, but on phase it was..
Hey skankinred I just also got Thermaltake Tide Water... I was wondering what temps your getting on thet 7900GT and also if your using heat spreades for the ram?
I have done rigerious tests on my 78000GT and it is now hooked up on my 7800GTX... I just wanted to see hows yours is running
I was actually able to boot into windows sine using your suggestions at 300x10 @ 133divider x3HTT 1.7vcore. x4HTT at 2.9 caused windows to freeze after 10 seconds.
Prime is still failing after 1 second.
Sledge, I can't going any higher than 551/843 since I haven't done the volt mod yet. I bought the resistor; can you tell me if this is the right one? The guy at radio shack didn't know if it was.
I have no idea on the resistor as I am not huge on modding things at this point in my PC life but I was able to get my 7800GT to peak at 495MHz Core and now have my 7800GTX peaking at 524Mhz Core clock... both run at about 38C to 40C under full load....
So I guess I am squeezing allot more out of my 7800GTX than the GT I haven't really pushed anything yet as I have not had any stability issues...
But I will be running test for the week back on the GT as I need to verufy the standard heat sink temps...
That is a variable resistor varrying from 0 to 50,000. I dont know what the resistance you want but if you want it within a 100ohm range or so you will want to get a smaller one. They should come in several 'sizes' at radioshack.
That is a variable resistor varrying from 0 to 50,000. I dont know what the resistance you want but if you want it within a 100ohm range or so you will want to get a smaller one. They should come in several 'sizes' at radioshack.
I need a 50k Variable resistor. Sounds like this is the right one.
Can you tell me which 2 of the 3 connectors i should solder wires to? The outside 2? It's going to replace another resistor.
I'm going to be doing this mod:
Do you think increasing LDT or Chipset would help any? Will it damage them if I raise the voltage?
For some reason I can pass Super PI 32m and 15mins of prime on the 2nd core @ 2.8 with 320x9@133 x3HTT, but prime ALWAYS fails on the 1st core in under 30secs. Is this a bad core holding me back?
How can the cores warm up when I'm using the Vapochill to freeze them? I was OC'ing to 2.5 for a week before I used my vapo.
Lemonlime, do you have the temperature sensor hooked into temp2 pin? I'm not sure if my temperature is right. Motherboard Monitor is showing my cpu @ 7C, Vapo says -39C at full load. Idle Motherboard Mon says -12C and Vapo says -47C. The hose doesn't feel very cold, only at the base right as it comes out the top of the vapo steel case.
Not warm up in the heat sence, but sometimes it takes the 2nd core sone time when OC'ing to catch up with the first core... take it slow and I bet your results will differ.
Lemonlime, do you have the temperature sensor hooked into temp2 pin? I'm not sure if my temperature is right. Motherboard Monitor is showing my cpu @ 7C, Vapo says -39C at full load. Idle Motherboard Mon says -12C and Vapo says -47C. The hose doesn't feel very cold, only at the base right as it comes out the top of the vapo steel case.
That sounds about right. The evaporator temperature will be much lower than the CPU core. The CPU temperature probe is never terribly accurate below zero, so the actual temperature is probably signficantly lower than what is being reported.
My CPU at load (before I messed up the heatspreader) was reading about +3'C at 3.2GHz, 1.58V and evaporator was -40'C or so. At idle, evaporator did about -53'C and CPU was about -26'C. I believe the temperature probe for the evaporator automatically uses pin 1, but make sure you don't hook anything up to probe #1. Did you buy the kit that gives you extra sensors?
The hose should not feel cold, or else you have an insulation problem. Have any pics of the setup? I'd love to see your system all setup with the vapo
Comments
This is as hard as I can seem to push my cpu on air.... It's only 51C. I'm at 1.425 vcore + 104% special vcore. If I go to 2.6ghz it's highly unstable. I'm not sure what preventing me from going higher.
What's the highest voltage I can safely run for operton 170's? I thought 1.45v is about the highest.
That can create a hard wall unless it is reduced at around 240MHz reference clock. Since you are at 250, I'd recommend the 4x HTT multiplier (sometimes listed at 1600/800MHz HTT in the bios)
I'm using 1.2v+136%special for 1.632vcore. I increased ldt to 1.3v. I can't even pass Prime at 2.9 with these settings. The revision on top of the IHS is CCBWE.
The vapochill says it's at -48C. Any ideas?
Did you get more info from the CPU stepping? Is it a CCBWE 0551 or 0550 etc?
In my experience, phase-change will buy you about 300-400MHz over air-cooling. Since you were stable at 2500MHz, I'd say a prime-stable 2800-2900MHz sounds about right for that chip. Could be that you are simply reaching the limits of the chip. My Opteron 148 did about 2.85GHz 100% stable on air, and about 3.2GHz totally stable on phase. You are lucky that you didn't get stopped short by the cold-bug too! Quite a few of these chips don't behave well at sub zero temps.
Could you take a quick A64Tweaker 0.6x screenshot for me? Perhaps we can find something holding you back there..
It's CCBWE 0543TPMW.
Actually, 1.63 is not that bad for phase. The maximum 24/7 voltage I'd be comfortable with on a 90nm AMD is about 1.65V actual. I've seen people do 1.65V on air/water too, and I know people have run chips for years without issues at that voltage. Even the vapoLS will start to choke a bit at higher than that. You can go a little higher, but I wouldn't recommend doing it for very long.
I'll see if I can find some info on that stepping..
I'm currently at 1.5vcore+110%special=1.65vcore. Vapo still says -47C. What's strange is prime runs for about 5mins on the 2nd core, but it always fails within 3 seconds on the 1st core.
I locked in memory timings @ 3448 in bios.
That all depends I guess. I can't say I've ever tried more than 1.7, but in my situation, the temps increased so much from 1.6 to 1.7 that it was pointless and actually worsened stability. Another thing to keep in mind is vcore stability. If your vcore jumps around at all, the danger is much greater. If you set 1.75 and it jumps around between 1.75 and 1.85, you could do some damage on one of those high spikes. I'm willing to venture a guess that you will not gain anything in stability by increasing above 1.65V..
It's generally accepted that 1.65V is about the 'safe max' so long as temps are kept in check. Any more than that is a risk, regardless of temperature.
About your stepping: I saw several examples of CCBWE 0543TPMW doing 2.6GHz with about 1.5V on air-cooling, so perhaps there is some more tweaking you can do to improve stability. Memory timings can make a world of difference (even when on a divider). Can you post an A64 tweaker screenshot?
That A64Tweaker shot looks bugged a bit, could you close CPU-Z and relaunch it?
1) Set trc = 9 and trfc = 16
2) Set twtr = 2, retry to see if Prime stable..
3) Set DRAM Drive strength = 7 and DRAM Data Drive = 2 and try Prime again
4) Set DRAM Response Time = Normal and try again.. Also try 'Fast' but avoid 'Fastest'.
5) Try to increase Max Async Latency = 9ns or 10ns
6) Try different Read Preamble settings, try to increase/decrease a notch or two.
I also remember reading on dfi-street on a tutorial with setting up the bios, they said to disable bank interleave. What's that for?
edit: When setting it to 133 it freezes after pci device listing as well. I have to use 140 to get past the bios to boot windows.
In regards to the early freezing in the BIOS at PCI device listing, I've seen that happen when the HTT is too high/low, or if the system is very unstable. Are you currently at the 3x HTT multi? Although it is not recommended, I sometimes find that using a higher than usual HTT works well on phase-change. 11x291 for me worked best with the 4x LDT multiplier for some reason. Wasn't the case on air, but on phase it was..
I have done rigerious tests on my 78000GT and it is now hooked up on my 7800GTX... I just wanted to see hows yours is running
Those settings look good dude, running out of ideas on my end. What is your max prime-stable setting? I'm guessing 2.8?
If you were only able to hit 2.5 on air, that could simply be all the chip's got I'm afraid. 2.8GHz dual core is still pretty awesome (FX62 speeds)
Prime is still failing after 1 second.
Sledge, I can't going any higher than 551/843 since I haven't done the volt mod yet. I bought the resistor; can you tell me if this is the right one? The guy at radio shack didn't know if it was.
So I guess I am squeezing allot more out of my 7800GTX than the GT I haven't really pushed anything yet as I have not had any stability issues...
But I will be running test for the week back on the GT as I need to verufy the standard heat sink temps...
Can you tell me which 2 of the 3 connectors i should solder wires to? The outside 2? It's going to replace another resistor.
I'm going to be doing this mod:
For some reason I can pass Super PI 32m and 15mins of prime on the 2nd core @ 2.8 with 320x9@133 x3HTT, but prime ALWAYS fails on the 1st core in under 30secs. Is this a bad core holding me back?
Maybe OC it to a comfortable level and slow increase it over the next week... instead of going full boar right away.
Lemonlime, do you have the temperature sensor hooked into temp2 pin? I'm not sure if my temperature is right. Motherboard Monitor is showing my cpu @ 7C, Vapo says -39C at full load. Idle Motherboard Mon says -12C and Vapo says -47C. The hose doesn't feel very cold, only at the base right as it comes out the top of the vapo steel case.
That sounds about right. The evaporator temperature will be much lower than the CPU core. The CPU temperature probe is never terribly accurate below zero, so the actual temperature is probably signficantly lower than what is being reported.
My CPU at load (before I messed up the heatspreader) was reading about +3'C at 3.2GHz, 1.58V and evaporator was -40'C or so. At idle, evaporator did about -53'C and CPU was about -26'C. I believe the temperature probe for the evaporator automatically uses pin 1, but make sure you don't hook anything up to probe #1. Did you buy the kit that gives you extra sensors?
The hose should not feel cold, or else you have an insulation problem. Have any pics of the setup? I'd love to see your system all setup with the vapo