Run it with AVX(win 7 SP1) otherwise is useless ;)
Printable View
Run it with AVX(win 7 SP1) otherwise is useless ;)
well it's not, I tested the newest 64 bit version against the newest version of linx with the newest linpack libraries and I could pass 7 + hours of prime and failed linx on the second pass.Quote:
Originally Posted by aoch88
this was with ht on, and linx set to run on 4 threads with SP1 installed.
There comes a point where you simply have to say "enough is enough." Just because everything except AVX with LinX passes doesn't mean that you're unstable. If every other normal program on the planet passes, including folding, except programs designed to abuse chips, then does that mean that people need to degrade their chips more, just to make that pass? If a program comes out that uses AVX and you're stable in it, I think you can just call it a day. Sometimes, you just have to know when to give. Unless your CPU is used only for benching and stress testing, of course.
What if the author of BurnBX came back and wrote a tester that heated the CPU 10C higher than AVX, and caused all the AVX stable systems to crash and burn? Does that suddenly mean you're not stable anymore? (if anyone even remembers BurnBX here, it would heat up P3's 15C hotter than Prime did in those days...I believe it used very short, extremely cache intensive programming loops, hitting the CPU *hard*. If anyone remembers that one Intel "Thermal Analysis Tool" that used to work with core 2 duos (but not yorks/wolfs) and the mobile versions, the torture test button in that did a similar thing temp wise, but was coded differently (and it was different than linpack).
I think it makes sense. I would just use Prime for an hour or two and if it passes, I'll start playing my games, etc. If no hangs, blue screens or whatever it is, that setting should be good enough. The problem is we have too many stress test around these days and to get all of them to pass might be a challenge sometimes.
stress testing sort of caters to an obsessive/compulsive part of the brain, where you feel you have to keep running over and over, to be satisfied that you're stable, and then you feel afraid that you got unstable after all that testing so you run AGAIN.
The bad part is, some of us ARE OCD--like me--and this just makes us break our hardware even more....
Well put Falkentyne, there still is NO ultimate stress tool... on X58 Prime blend ( 6 hours ) and LinX 50 loop stable... yet Hyperpi 32m ( 12 threads) freezes after loop 2 , Bad company kicks me to desktop,....lol.... daily stability is what you need, not some silly Gflop screenshot...
Seems like many MC stores don't follow through with what they claim on their site, which says, "Only one CPU per customer..." but I do envy those that live by stores that don't care about how many CPUs I buy at a time. After all, aren't they in business to sell? Then why limit it? If a customer wants to buy ALL the CPUs in one visit, sell them all at once! It's very stupid to say, "Oh, we are here to sell, but only sell one at a time so the party can last longer. We don't want to sell fast. We only want to sell slow, one at a time." Afterall, more sale just means more business for the company. I think this policy is very stupid.Quote:
Originally Posted by l0ud_sil3nc3
Guys, Im really trying to get any nice high clock to be 3D stable, I did get this far in 2D .... any good?
pic coming .....
:up:
Any idea in what a hell is going on ?
http://img713.imageshack.us/img713/8034/linxfailn.jpg
Actual load vCore was 1.368v
Probably means you need to increase the vcore about 0.015 notches.
Though frankly why are you running so many passes?
And that's a great cpu. Stop torturing it with LinX AVA or you'll see that vcore rise. Sorry if I asked this before (I asked a few people), but what's the default VID shown in the BIOS (usually under "target") at 3.4 ghz, and at 5 ghz? I'm curious.
I
Crazy minimum VOLTAGE? XDQuote:
Originally Posted by st0ned
Just saw he had HT off. Would be a lot more interesting with HT on.
I might be able to do AVX on my chip with HT off too, with around 1.4v load vcore.
vCores is 1.368v not the 1.39X show by CPU-Z, I'm only using LCC at 75% that gives me that drop between load and idle, but prevent me from getting random BSOD's on normal daily tasks.Quote:
Originally Posted by Falkentyne
It run's 5.1 with only 1.416v .
HT with 1st SB linpacks ~80Gflops were verry easy to pass. With new linpack ~110Gflops they got a bit harder, but It will do it around 1.38v I guess :)
Anyone know how to determine the true current clock speed with any of the monitoring apps? (speedstep, c1e, etc. enabled)? CPU-Z, RealTemp, CoreTemp, generally disagree with each other. CPU-Z is farthest off from the others, usually displays minimum or maximum.
EDIT: In light of my page-reloading failure, I am editing to include my thoughts on the stress testing bit.
I found starting with LinX AVX to be the best bet, and it was the most likely to pass 20 runs for me. 3 times I passed on what I thought was a stable clock and then failed prime blend in <1 hr.
Now my stress testing goes as follows:
LinX AVX 20 pass * --> prime small fft 4hr** --> prime blend 8hr --> 3dmark11 extreme 2-4hr
*use as much ram as you can w/o swapping (14gb for me)
**impatient, run longer if you like
You would be surprised what passes LinX AVX and then gets destroyed by small fft or blend :(.
-sfxs
Hello !
I got a 2600K today with batch 3101A359...*A0645. I will test today and post results here but... Initially, this is good or bad batch ???
Sorry by my bad English.
Thanks
No one knows,test it and tell us :)
:D Please now agree with me
damn Intel give us fully unlocked cpu and not half way.. we demand atleast X70 of multis
:eek::woot:
Out of curiosity, anybody noticed when running say IBT with Core Temp that CT will at random points change VID? Anyone figured out if there is a way to stop the CPU from requesting a high VID? My 2500k is 100% stable with 1.35v @ 4.6GHz but when the damn CPU requests 1.4v VID it makes IBT fail because the CPU gets too hot (1.35v 65c, 1.4v 80c!) :brick:
The only way you can do that is with manual volts. VID is sort of the asked voltage for that Mhz and workload, so if you push him more at the same Mhz the VID will also go up.
^^ what he said.
And Stoned, did you miss my first post above? You replied to the second one though.
I was asking,
what is the default VID of your golden CPU, when it is set to X34 in the BIOS? The vid is to the left of the voltage selection area, (not the health monitoring vcore), and the VID is what the CPU will select for voltage if it were set to auto volts (example VID could show 1.21v even though you set the vcore manually to 1.3v, for instance :)
Also, what is the default VID showing when you set it to 5 ghz?
I'm just curious about your chip.
Thanks.
For the 5G is what it was in the PS 1.37v I think, but when loaded he will ask for a bit more 1.39(?). Default VID is high nevertheless around 1.25v. This as the voltages shown by CT, but in the bios with multi 34, it's around 1.23v
Thats weird then. I have voltages and LLC manually set and the CPU still does it in my case. Time to fiddle with a few things I guess. I'll see if outright disabling speedstep features and so forth makes any difference.Quote:
Originally Posted by st0ned
Just upgraded my 7 to SP1 but looks like AVX still wont work.. how can i get it work??
Current setting is totally Linx stable, but havent tested it with AVX..
http://i6.aijaa.com/b/00540/7880581.png
You also need to DL the neww Linpack from Intel website and replace the files in LinX Folder with them.
Hey, It wont make you VID stop floating, mine also floats, but it will prevent you vCore from floating with it :) . At least it should...Quote:
Originally Posted by Ket