Ya when I get a Phenom think I'll skip overdrive unless I see you guys reporting it works great.
How are you measuring W AC?
Ya when I get a Phenom think I'll skip overdrive unless I see you guys reporting it works great.
How are you measuring W AC?
"Angels on the sideline, Puzzled and amused.
Why did Father give these humans free will?
Now they’re all confused."
9850(JAAFB AA 0810BPAW), DFI 790FX M2RS(3/05 Bios), G.SKILL 2 x 2GB 1T-5-5-5-15, OCZ 600, MINI NINJA, WD SE16 AAKS 320, HIS3850, bluegears b-Enspirer, Vista 64
With an power meter for domestic appliances.
I already had the HR-05 already from my old system so I was planning on just swapping it over. Besides, the small copper ramsinks can't keep a northbridge cool.Originally Posted by ZL1Killa
The Cardboard Master Crunch with us, the XS WCG team
Intel Core i7 2600k @ 4.5GHz, 16GB DDR3-1600, Radeon 7950 @ 1000/1250, Win 10 Pro x64
That's quite a lot. It can't be unless something is placing a load on the CPU or some voltage is high. Even if the NB is under load. Have you benchmarked EVEREST to compare L3 cache values at the same speed with bw before this phenomena?
Yes, I was going to write you a customized file to try that's why I asked about voltages and VIDs.No it has only a voltage option for cpu and nb, no extra vid's. Those stay at 1.25V no matter what vcore is set (have not tried cnq).
Also modified the DefaultConfig.xml file but i might have modified the wrong voltage there. An alternative whould be using an customized profile whom get's loaded at start.
Can AOD change VIDs on your system?
I had planned to pick up the Sapphire as the next RD790 board but when you say it doesn't have VID options, that scares me.
Try this MSR: 0xC0010070
EDX = 0x00000000 (always)
EAX
0x3000400A = 1.150VID
0x3000385A = 1.200VID
0x3000375A = 1.212VID
0x3000350A = 1.225VID
0x3000300A = 1.250VID
0x3000285A = 1.300VID
0x3000275A = 1.312VID
0x3000250A = 1.325VID
0x3000200A = 1.350VID
It should change CPU VID and hence volts (check in AMD PM).
I'm not sure of all the ranges because I've not played with it more but the pattern is obvious. It's very easy to change VID and voltage this way without problems (you can include it in RWE).
Yeah it's really weird. Had replaced the stock cooler with the big typhoon and the ram from gskill to crucials few hours before that, switched back but that did not make a difference.
NP i'm familar with xml-files.Yes, I was going to write you a customized file to try that's why I asked about voltages and VIDs.
Have neighter tried AOD nor WPCREDIT to change em, just noticed that they stay at 1.25V in AMD PM no mater what vcore an vcorenb i choose.Can AOD change VIDs on your system?
I had planned to pick up the Sapphire as the next RD790 board but when you say it doesn't have VID options, that scares me.
Try this MSR: 0xC0010070
EDX = 0x00000000 (always)
EAX
0x3000400A = 1.150VID
0x3000385A = 1.200VID
0x3000375A = 1.212VID
0x3000350A = 1.225VID
0x3000300A = 1.250VID
0x3000285A = 1.300VID
0x3000275A = 1.312VID
0x3000250A = 1.325VID
0x3000200A = 1.350VID
It should change CPU VID and hence volts (check in AMD PM).
I'm not sure of all the ranges because I've not played with it more but the pattern is obvious. It's very easy to change VID and voltage this way without problems (you can include it in RWE).
Currently running an stability test with 219x12 will try the MSR way tomorrow and let you know if it works.
Has anybody gotten this board working with G.Skill 2x2GB 1066, or TeamGroup 2x2 1066? The G.Skill yields a C1, and the Team shuts down the board straight from Memtest after about 20 seconds of Memtest.
Haha thats what i did it dropped my temp ~2-3C and is roughly @ 40-42C under full load.
I also have a 80mm fan blowing at the PMW heatsink.
Last edited by Jaivan; 02-26-2008 at 11:54 PM.
Achim: I rebooted at the same settings and my power draw idling is now down to 107W AC again for about 3 hours.
I am pretty sure it was IMC related because IMC performance is now also better.
Mine primed fine for 13hrs with the higher power consumption. Upped the ref HT for another MHz and the vcore to 1,39325V and testing 220x12 atm. The consumption in idle is still 30W higher. Decided to go till 222 cuz i'd like to have ~2,66GHz to compare it with the intel rig.Achim: I rebooted at the same settings and my power draw idling is now down to 107W AC again for about 3 hours.
I am pretty sure it was IMC related because IMC performance is now also better.
Mine's now idling 30W AC lower and I'm slowly dropping the volts. At 1.272V for 2.613GHz for over a day and night to going to drop it lower. Stock all else volts.
Hmm I removed all additional stuff (fans,cdrom...) tried different slots for the ram and still had the +30W. Tried the 3850 from the intel rig and different pci-e slots without luck.
Switched to the BE2400 and verified the same power consumption at 3.2GHz i had few weeks ago.
After switching to the 9600be the power consumption is now at 148W AC at 2.6GHz (13x200) running at 1,35V now, guess i can go lower.
So the issue is gone now. Puh, was 199W at 2.6GHz before.
Above 2.628MHz system is not stable with vcore below 1.4V, dont want to use higer vcore atm.
Trying 1.28V vcore for 2.6GHz now above the voltage requirements rise very quick.
I think i can reproduce something comparable to your idle freezing. If i run 3dmark06 with prime the system freezes if i submit the 3dmark results. That additional test shortens testing a little.
The freezing looks like it's a board/BIOS issue because people on other boards aren't getting it. It occurs with a fully +5-20 day 24hr Prime95, fully benchable setting. Nothing will make it error until you either play a Youtube video/SWF/Flash Game/Firefox or just idle it for hours randomly. At that moment power consumption jumps to around 80% max and CPU freezes (with or without the TLB errata patch). Some time you get it after 10 minutes uptime and at other times you get it after 3 days uptime. Very random.
NM I'll have your board soon. What's best BIOS you're using? Have you tested BIOSes and their perf? Which one has TLB patch and which ones not?
No chance for such a long uptime here in the next two weeks, due to the amd intel comparison i plan.The freezing looks like it's a board/BIOS issue because people on other boards aren't getting it. It occurs with a fully +5-20 day 24hr Prime95, fully benchable setting. Nothing will make it error until you either play a Youtube video/SWF/Flash Game/Firefox or just idle it for hours randomly. At that moment power consumption jumps to around 80% max and CPU freezes (with or without the TLB errata patch). Some time you get it after 10 minutes uptime and at other times you get it after 3 days uptime. Very random.
Thank you for clarifying the freezing issue, I thought it occured repeatedly after ~3days
Better get your own.
Used the latest sapphire provided first, switched to the latest stable dfi bios (22/01/08) after the 9500 stopped working. Brother Esau posted the flags required to force the flash somewhere here in this thread.
I did not do much beside trying to reach the cpu frequ limit the last weeks, that 3hrs freeze made that a long term task.
The bioses previous to 22/01/08 had all types of glitches with phenoms, so i did not yet try em.
You can enable/disable the tlb cache via bios. I currently check what MSR registers are affected. Will run the everest mem and cache benchmark we ran on the msi and asus mobo and add results here.
exactly. also for the NB.. i wouldnt' think a hr-05 would cool it either, but i will look into aftermarket copper heatsinks to see about mounting some there or something
the cpl-3-150 chips are good until 125C..which i wouldn't ever want to get tooo but thats the highest they can operate.
whats the max temps for the NB /SB? by the chip manufacturer?
2013 Rig: Mythlogic Pollux 1713: i7 4800, 16GB 14900 Corsair, AMD8970, 840 Pro RAID 0 (128gb), 265GB 840 Pro, 1TB HDD, Killer 1202, Win 7 x64
PREVIOUS Desktop:EVGA E760 3-Way, 2x 5870, Corsair GT 6gb 1600, i7-920 @ 4.35ghz 1.42v, GTZ, EVGA waterblock, 2xthrmchil120.3+mcp655, Silverstone DA1kW, ARC-1220
Laptop- ASUS G50V-A2@3GHz w/2gb RAM G.Skill
Car: 01 T/A WS6 http://www.ls1tech.com/forums/showthread.php?t=631954
...Ur car is sneezing?!...
Help Save Lives Join World Community Grid!
TTTTTTTTTTTTTTTTThhhhhhhhhhhhhhhhaaaaaaaaaaaaaaaaa aaaaaannnnnnnnnnnnnkkkkkkkkkkkkkkkkkssssssssssssss sss AAAAAAAAAAAAAAAAAaaaaaaaaaaaaccccccccccccccccccccc ccchhhhhhhhhhhhhhhhhiiiiiiiiiiiiiiiimmmmmmmmmmmmmm mmm!!!!!!!!!!!!!!!!!!!!!!!!!!!!
BBBBBBBBoooooooooooooooooaaaaaaaaaaaaaaaaaaaaaaaar rrrrrrrrrrrrrrrrddddddddddddddddd UUUUUUUUUUUUUSSSSSSSSSSSSSSSSSsssssssssssssssbbbbb bbbbbbbb iiiiiiiiiiiiiiiiiisssssssssssssssssssssssssuuuuuuu uuuuuuuuuuuuueeeeeeeeeeeeeeeesssssssssss.......... .........
Cache and Memory Benchmark (TLB Cache enabled via bios, should akt like yellow mode).
M3A results
K9A2 Platinum results
Memory latency is better than on the M3A and comparable to the K9A2 P. All other results are slightly lower.
Checked the TLB fix, both registers are changed on all cores, so the fix is waterproof.
Also tried to modify the CPU VID. AMD PM reports the change but the power consumption does not change when I switching between 1.3V and 1.225V.
KTE, didn't you mention that vid's are not needed if the core voltages can be manipulated?
BTW: With those timings the 1T setting is not applied (tried Auto/enabled/disabled) on boot but can be changed with Memset. 1T/2T could not be modified on the M3A.
Last edited by justapost; 02-27-2008 at 07:44 AM.
USB problem? Go on with smilies.TTTTTTTTTTTTTTTTThhhhhhhhhhhhhhhhaaaaaaaaaaaaaaaaa aaaaaannnnnnnnnnnnnkkkkkkkkkkkkkkkkkssssssssssssss sss AAAAAAAAAAAAAAAAAaaaaaaaaaaaaccccccccccccccccccccc ccchhhhhhhhhhhhhhhhhiiiiiiiiiiiiiiiimmmmmmmmmmmmmm mmm!!!!!!!!!!!!!!!!!!!!!!!!!!!!
BBBBBBBBoooooooooooooooooaaaaaaaaaaaaaaaaaaaaaaaar rrrrrrrrrrrrrrrrddddddddddddddddd UUUUUUUUUUUUUSSSSSSSSSSSSSSSSSsssssssssssssssbbbbb bbbbbbbb iiiiiiiiiiiiiiiiiisssssssssssssssssssssssssuuuuuuu uuuuuuuuuuuuueeeeeeeeeeeeeeeesssssssssss.......... .........
OT: Lol, had to cmos flash the p5k-e yesterday cuz it did neighter recognize and usb nor an ps/2 keyboard.
Update: Damn having the power consumption issue again after i switched back to the 2.6GHz settings. It are 182W AC in idle now, thats +40W.
Last edited by justapost; 02-27-2008 at 07:53 AM.
Oh really... Interesting. My system froze randomly on me a few times (this is all at stock speeds right now mind you), two times it was in firefox. System had been running FAH SMP no problems. I upped my ram voltage (rated at 2.0-2.1V, bios had it at 1.9V) so we'll see what happens with that but so far no freezes since.The freezing looks like it's a board/BIOS issue because people on other boards aren't getting it. It occurs with a fully +5-20 day 24hr Prime95, fully benchable setting. Nothing will make it error until you either play a Youtube video/SWF/Flash Game/Firefox or just idle it for hours randomly. At that moment power consumption jumps to around 80% max and CPU freezes (with or without the TLB errata patch). Some time you get it after 10 minutes uptime and at other times you get it after 3 days uptime. Very random.
BTW, I forget which is better for speed/overclocking, ganged or unganged mode?
If that stock chunk of aluminum on the NB can keep it cool enough at stock speeds surely the HR-05 can keep it cool at overclocked speeds. Maybe need a fan at really high speeds but then again maybe not, we'll have to see. I'm thinking of fabbing up a custom bracket to mount mine with.
Last edited by Sparky; 02-27-2008 at 10:04 AM.
The Cardboard Master Crunch with us, the XS WCG team
Intel Core i7 2600k @ 4.5GHz, 16GB DDR3-1600, Radeon 7950 @ 1000/1250, Win 10 Pro x64
Ganged mode means both dct's act as one 128bit-dct. It's faster for single threaded stuff.
Unganged means both dct's operate independant, tha's better for multitasking. But that mode is harder to oc.
Sandra and RightMark Multithreaded Memtest are two examples whom profit from unganged mode.
Those two screenies i posted are ganged vs. unganged singlethreaded.
Idle power consumption is dancing on my nose today. Had to remove one HD for an other machine, so i had to reinstall. After that i had 150W again. Now comes the funny stuff. I plugged in my USB-Stick with Benchmarks and Utility stuff and voila back at 180W. That USB-Sticky requires around 2-3W more if i plug it into the m3a or m2a-vm.
Last edited by justapost; 02-27-2008 at 11:17 AM.
Thanks for the results!
Interesting, the board is slower overall, especially in L2 and L3 and latencies. The Plat. benches a little higher than that as I bench with all software running to show ideal bandwidth there (AV/FW in background just paused). It will fall shorter of it which I didn't expect.
Yeah the TLB fix isn't applied there you can see by your L3 scores. If it is, the L3 is very low and slow or it doesn't run at all.Checked the TLB fix, both registers are changed on all cores, so the fix is waterproof.
BTW does Memset not work fully for you? It was displaying wrong speeds there for some reason.
It's the correct register as I've seen in the guides, try 1.55VID and see if idle power shoots up.Also tried to modify the CPU VID. AMD PM reports the change but the power consumption does not change when I switching between 1.3V and 1.225V.
BTW, what does MSR C001_0064 show, do you remember? I've forgotten, it's something related to the NB IIRC.
True, they're only to allow you more intermediate, high and low volts, nothing else. So if you want a voltage between say 1.225V to 1.23V and your BIOS doesn't allow it, then changing VIDs will give you an option for that voltage in there.KTE, didn't you mention that vid's are not needed if the core voltages can be manipulated?
Yea USB problems on this board and they crap out like that randomly. I was stuck because you can't press any key at all so you can't shut down but a power supply off while OS is running which I didn't want to do as it destroys system too much and I had many works unsaved.USB problem? Go on with smilies.
OT: Lol, had to cmos flash the p5k-e yesterday cuz it did neighter recognize and usb nor an ps/2 keyboard.
That's a lot for idle. Somethings wrong somewhere. It could be your PSU too, if it falls in efficiency steeply.Update: Damn having the power consumption issue again after i switched back to the 2.6GHz settings. It are 182W AC in idle now, thats +40W.
Checked to make sure all MSR's are modified.Thanks for the results!
Interesting, the board is slower overall, especially in L2 and L3 and latencies. The Plat. benches a little higher than that as I bench with all software running to show ideal bandwidth there (AV/FW in background just paused). It will fall shorter of it which I didn't expect.
Yeah the TLB fix isn't applied there you can see by your L3 scores. If it is, the L3 is very low and slow or it doesn't run at all.
Did you run your benchmarks with weak or normal mode for the mem?
Ups, the readings have spikes, catched one by accident.
Now that cpu should survive till k10.5 at least, so i tried 1.1VID at 2.6GHz. I expected prime95 to fail but it did not also no power consumption change.It's the correct register as I've seen in the guides, try 1.55VID and see if idle power shoots up.
In case you missed cuz i updated an old post, i have an USB-stick requiring 30W.
C001_ 0064h-0068h are the P-State status registers, first bit is interesting.
It's zero here, that means the rest of this msr register is invalid.
You can modify vcore in steps of 0.00625V that is half of what you can do with VID's (0.0125V step size).
Don't like USB keyboards, still causing issues i don't want to spend my time fixing. Good to have a vnc-service running in such a case.
This board tried to kill my processor, ran 1.63 volts through the core by default. I definitely didn't tell it to do that. It's currently running old Ballistix 2x1, but I'd bet the Team 2x2 starts working now that the processor isn't hitting thermal limits.
This DFI doesn't like the TeamGroup RAM.Try it with SuperTalent or Crucial.
What do you mean by weak and normal mode here?
Hmm.. it modifies them on all Phenoms according to guide AFAIK. What voltage and VID did it give you?Now that cpu should survive till k10.5 at least, so i tried 1.1VID at 2.6GHz. I expected prime95 to fail but it did not also no power consumption change.
Try 0x3000100A from 1.25VID, it should give you a ~10W AC idling increase in power. If not then something with your CPU MSR is going iffy.
Yeah I saw it, it's very weird and well, too highIn case you missed cuz i updated an old post, i have an USB-stick requiring 30W.
Thanks but are you sure that's their only relevance?C001_ 0064h-0068h are the P-State status registers, first bit is interesting.
It's zero here, that means the rest of this msr register is invalid.
Have you actually tried these work? Very good options if they do. What (%) Vdrop and Vdroop does the board have?You can modify vcore in steps of 0.00625V that is half of what you can do with VID's (0.0125V step size).
Tried both, last time it was PS/2, both have same problems here.Don't like USB keyboards, still causing issues i don't want to spend my time fixing. Good to have a vnc-service running in such a case.
DDR2 800 4-4-4-12 2.16V as of last night.
Posting Permissions
Reply With Quote
Bookmarks