Tried it yesterday, and the best side of it is that after benching you can just peel it off easily. :toast:Quote:
Originally Posted by mikeguava
Printable View
Tried it yesterday, and the best side of it is that after benching you can just peel it off easily. :toast:Quote:
Originally Posted by mikeguava
Well, I booted up with the naked cpu last night and there seems to be a few issues....First, my Win2k3 install refuses to work properly with this Opty. I've tried everything - including re-installing 4 times the last 2 days - and nothing seems to let the cpu run stable. I barely can complete any benches @ high speeds, however, I ran an 8m @ 13x270 and the result was 27+minutes:confused: I'll get the screenie posted later.
The other issue has to do with temps. It seems as if removing the spreader did next to nothing in terms of improving temps on the cpu. In fact, the temps seem to be a little worse now that the spreader is off. I checked the mount after the bench session and it was absolutely perfect. What makes this even more odd is that my evap temps continued to remain below -70C loaded while the core temps reached as high as -18. I'm hoping there's not some sort of "clog" in my evap head. I'll talk to Chilly1 later and see if he may know what is wrong here.....
Man steve that really sucks... Have you tested the cpu with a reg. heatsink to make sure it didn't get hurt when poping the hood (doubtful but maybe..) I really wanna see this baby break some records!!!!!
Oh and for insulating the resistors under the IHS I use nail polish...works really well, and can take it of with remover and looks like new!
Good luck man lets us know!
Happened to me once, with a container and a venice 3200+. After removing the IHS temps went up from -35º to -26º... Mounted it again like three times, but always the same... Odd, really odd...
I havent tried my 154 in subzero cooling yet, but with my watercooling system (which holds my Venice 3000 @ 3Ghz 1.7v @ 38ºC full) the 154 gets pretty hot (around 51ºC full @ 1.55v).
If you are saying that removing the IHS doesnt help at all, then I will rethink if I want to remove mine...:stick: :stick:
Hey Lord! :toast:Quote:
Originally Posted by Lord_RTKK
Did you check the mounting? That's damn hot for water...
Yeah man, the mounting was perfect, I checked it twice or three times... but those 154 get pretty hot... :stick:
My WCing setup is well designed and equilibrated (see my last post to see my vennie temps ;)) but this 154 gets hot like a demon...
I thought that it could be because of a bad contact between IHS and core, but now I am reading stephen´s experience with his 154, I am doubting about removing IHS or not...
Next week we will get the 154 under DICE, so we will se if this heat helps to avoid cold bug out... :D
Large System Cache on?Quote:
Originally Posted by s7e9h3n
It was, but for some reason Disable Executive Paging was off and Second Level Data Cache was set to 0. Looks as if the registry on my 2003 install is totally borked. Send me a copy of yours ;)Quote:
Originally Posted by freecableguy
UPDATE: I've got the cpu finally figured out!!! :woot: ATM, it's running 32m @ 3700, 1.61V.....idle @ -47, load @ -34........Stay tuned as I'll have some benches finished by tonight
At last :slap: :D :toast: Don't even show up without 19:xx :lol: Good Luck, Steven :up:Quote:
Originally Posted by s7e9h3n
Nice to hear, but what was the problem then, always handy to know if someone encounters same problem in the future :)Quote:
Originally Posted by s7e9h3n
I think it may have to do with the tremendous heatload the 154 creates. For one reason or another, the evap on my autocascade wasn't able to handle such a concentrated heatload from the core and as a result, it basically became "overloaded". The solution? I installed my trusty little cold-plate I normally use to RAISE temps on cold-bugged cpus:Quote:
Originally Posted by DaWaN
http://www.cryo-laboratory.com/uploa.../coldplate.jpg
Here's what was happening before installing the cold-plate :rolleyes: (this must be some kinda record):
http://www.cryo-laboratory.com/uploa...3x2758mwtf.jpg
And this is what happend afterwards :cool: (No tweaks at all, just a couple of RAM timing changes):
Sorry Michal - I'm workin on the 19:xx as we speak ;)
Spi32m, 14x264, 1.61Vcore
http://www.cryo-laboratory.com/uploa.../14x26432m.jpg
Spi8m, 14x268, 1.61Vcore
http://www.cryo-laboratory.com/uploa...4/14x2688m.jpg
Spi1m, 14x270, 1.61Vcore
http://www.cryo-laboratory.com/uploa...4/14x2701m.jpg
MUCH, MUCH MORE TO COME ;) ......
Finally! Very nice runs...
Very Nice results s7e9h3n. I guess this is with your team group PC 4000?? Does the 154 has stock unlocked multipliers? Was that 32mb ran 1:1?
No unlocked multiplier on the 154- It's not an FX after all. Hope to see more spi:stick:
Under the plating the IHS is a solid copper slug. I'm guessing, but what I think would happen in a sub-zero situation with the IHS on is that you'd have a cold-sink sitting next to the core, so when the core heats up you're able to transfer the heat into the IHS and warm it - this would draw heat out of the core very quickly, I think. Once the IHS is off you no longer are drawing heat into the copper slug but rather into the evap head. Now, I'm not a phase guy but I'm going to guess that the phase head might not be able to sink heat as quickly as the copper IHS.Quote:
Originally Posted by PoL
I used to run a tec to cool my celeron 333a (550mhz for the win) and I had much better luck (and core temps) using a block of aluminum between the core and the TEC than I did just using the TEC directly on the core.
edit: are people seeing better temps on phase with the IHS on, or off? or is it a mixed result?
I had the same thing happen to my 146 steve, IHS off and temps are same if not slightly worse on water.
Cant compare with DICE though as i dont use it as cold as possible due to bug at -25C.
I havent attempted to solve this issue, but a thin copper plate might be the answer.
A thin copper plate may help in the sense that it will help keep load temps down, but the flipside to this is that it also takes much more capacity from the cooler in order to displace the amount of heat generated from the cpu and the copper itself.....Quote:
Originally Posted by T_M
Nice one you got there Steve:toast: ..Its a gem:)
154 is a hot chip...Heres with single stage load @.......
http://img17.imageshack.us/img17/566...shot1285eg.jpg
My ss cried....evap went positive:p:
Positive Evap???? Holy $hit :eek: How's this for -65C Evap LOADED ;) ?:Quote:
Originally Posted by Dumo
http://www.cryo-laboratory.com/uploa.../14x26832m.jpg
Those Opties are geting better and better when time passes......:)
Good job s7e9h3n.....:) :toast:
I know everytime I see a new Opty they are getting better but this one is waaaaay better. They are also getting hotter and hotter...Quote:
Originally Posted by hipro5
Thanks....but this cpu can only dream of the spi times your 4Ghz Dothan can do :toast: Now please show me a mod to give me the x15 multi since I'm hitting a brick wall @ 14x270 :p:Quote:
Originally Posted by hipro5
BTW, this cpu is FAR from a new stepping. In fact, it's a sample from an earlier stepping which it shared with the CABCE FX57's. Sadly, the newer steppings can't touch these results due to the cold-bug.....:(
Im convinced, the hotter the chip the less chance of it having cold bug. As long as you have the cooling in place to handle the heat load (which you have) these chips can keep going and going, well ofcourse not to affinity, hehehee
Thanks for sharing the information on how you solved this, UTKightmare has been having similar issues with the 0530APMW i sold him, it gets alot better temps with the IHS on then with off and it is also a hot chip. Im hoping he can get some ncie cooling sorted soon and with the info u shared hopefully there will be some more competition.
:toast:
Nive vcore Steve!! :toast:
Now, you're talking Steven :) Great clocks and fast run :up: I'm glad you finally figured this CPU out...Quote:
Originally Posted by s7e9h3n
I wonder If any of the lower models (146-150) will ever touch that speed... I bet not :(
Keep pushing it... Who knows, maybe 3.8GHz in 32M is doable :toast:
Thx...atm, it doesn't look likely that a 3.8Ghz 32m is possible because I'm hitting a wall @ 14x270. And unless my attempts at unlocking the cpu by switching memory and setting weird values from bios works :p: , it won't happen. The cpu looks as if it has a cold-bug of sorts at these temps. It can't sustain a Spi run for long @ these HTT's. I've got to brainstorm and see what I can figure out.......Quote:
Originally Posted by bachus_anonym
TBH, I really don't think that ANY other opty that we'll see will be as strong as this cpu. Quite frankly, this particular chip was strong enough to be binned as an FX57, but became one of the initial cpu's in the Opteron line........
Yes I know.......My DOTHAN is made in year 2004....;)Quote:
Originally Posted by s7e9h3n
As a matter of fact some minutes ago, I asked esdee to bring me tommorow the dataseets of the A64 CPUs......I think I have found a way to push them a bit higher with a mod on any mobo for them.......;)
I have an 770 DOTHAN that could do with the default adapter/cooler about 2850MHz.....After moding the adapter, I was able to push it at 2940MHz.......Today I did some more mods on it which I think they will be almost identical with an A64 platform......With the new mods onto the adapter now my 770 works at 3043MHz with the default heatsink as for Super-Pi and 3030MHz 3D benchable.......;)
Make sure you keep all of us posted :toast:Quote:
Originally Posted by hipro5
But don't forget one HUGE difference here - the reason why my cpu won't go higher is not because it's not capable of it, it's because the cold temps don't allow it. So in order for me to reach a higher frequency, I'd either have to gain access to a higher multiplier or defeat the cold-bug so I can run higher HTT's. Dothan's don't usually have any cold issues @ -50C core temps :p:
This locked multi is kicking my ass :p: I've found the exact point where this cpu starts freaking out - 272HTT @ -73.8C Evap. Once temps fall down that low, everything locks up. This is making it VERY tough to get off any SPI runs above 3800. I did manage to run off an 8m (actually quite a few) but 1m was another story. Once the bench finished, I literally had ~5 seconds to open all the necessary windows (cpuz, smart guardian, a64 tweaker, irfanview) before I captured the screen and saved it. It took me about 15-20 tries to finally get the screen saved before my rig locked up - but I must say....it definately was worth it :woot: Here's the results of today's work :p: :
http://www.cryo-laboratory.com/uploa...4/14x2708m.jpg
And this is the one that took quite a few tries to get saved:
http://www.cryo-laboratory.com/uploa...4/14x2721m.jpg
I'm about to try to corrupt my bios on purpose in hopes of unlocking the multipliers on this cpu :rolleyes: If I had access to the x15 multiplier, a 1m @ 3900 would surely be possible.....
why not loosen MAL/preamble for more stability ??
The cpu is rock solid at these timings...Actually, I've been able to finish those benches with RP as low as 4. It's just that this combination returned better results than any others. Tried them all (within reason): MAL 8 and 9, RP 4, 4.5, 5.5, 6 - all ended up with slower times. The problem isn't with stability at these speeds, rather it's a problem with the cold-bug ;)Quote:
Originally Posted by eva2000
ah pesky cold bug i see :)
Why don't you try "multiprocessor driver", Steven?
That's possible?Quote:
Originally Posted by s7e9h3n
While it is possible to corrupt your bios on purpose - I was just kidding about trying it :p:Quote:
Originally Posted by PoL
@ Bachus: I just may try that ACPI Multi Proc Tweak - especially if I'm gonna run off some 3dMarks tonight ;) Oh, BTW, time for you to do some work to get back into the top 20 on Holicho's 32m ranking. Sorry, my score knocked you out. I'm sure we'll see you back up there again soon though.....:toast:
haha :DQuote:
Originally Posted by s7e9h3n
If it was true tine_off would go crazy. :D
Good! Hopefully it will help you out keeping that Opty hot enough :)Quote:
Originally Posted by s7e9h3n
As for Holicho's Ranking, I was surprised that you've never sent him anything before... Nice debut :) As for me coming back to Top20... I don't think it's gonna be that easy. I would need at least 10s (preferably 20s) and somehow I don't see it to be possible with cheap Optys, like mine. Despite your 154 being called Opteron I don't think there's any other Opty that will match yours. I'm pretty sure that Opteron 154 has more of "FX quality" than "Opteron quality" in there... Man, and to think that you're still complaining about lack of x15 multi :D :p: You're already very lucky to have x14 :slap::lol:
Grats on a super SPi32M, once again Steven :toast:
Woah, I didn't notice the difference between you and the next guy up. Look like you'd need another 100mhz on the cpu. Now if you can pull that off, you'd DESERVE to be in the top 10 regardless what your score is :p: The whole reason I was attracted to this Opty in the first place was because it didn't looik like the typical Opteron. But it definately shares one major similarity with all the other Opties - this thing puts out a TON of heat. Nothing I've ever threw under the autocascade has pushed the unit like this cpu has (and yes, that includes an x2 @ 1.67V). I'd say the heat coming from this cpu would be at LEAST 1.5x that of an FX at similar Vcores. I'm still trying to figure out how it would be even possible to run this cpu on anything other than high end water or better.....:confused: Thanks again MichalQuote:
Originally Posted by bachus_anonym
Sometimes I think that IHS seriously criples those Opterons (or maybe SDs in general) not some sort internal differences between them and other chips. I have no idea how to explain it, though... Just look at those screenshots below and note LOADed temps with my Opteron 146 - before and after removing IHS:
Opteron 146 @3400MHz, IHS ON, SmallFFT .:. VS .:. Opteron 146 @ 3500MHz, IHS OFF, Small FFT
I literally don't know what is up with that... QC issue or sth :shrug:
S7eph3n, try running SP2004 in the background, set at lowest priority...you can mozey on around geettin that therr programs up.
Anyway, this chip is a beast for sure....but wouldn't a divider help a tad? SNIP EDIT: nvm, that made no sense....have you tried a divider though? knock down the mem controller speed a few notches....
And B_A...HOLY TIHS! Was the IHS even making contact?
Got a pic of the thermal paste underneath the IHS?Quote:
Originally Posted by bachus_anonym
Quote:
Originally Posted by Vapor
Yep!... One thing I noticed was that I had to literally scrub it off of IHS as it was dead dry...Quote:
Originally Posted by s7e9h3n
It was still clocking high, even on air. I've seen original owner's screenshots on regular air and it did SP2004 at 3.0xGHz, 1.4v. But temps were around 50C LOAD...
EDIT: Anyway... I don't want to hijack this thread :) Let us know how "multiprocessor driver" works out for you as soon as you try it ;)
I'd have to say, that's a lot better contact patch than this 154 :rolleyes: 50C loaded temps wouldn't be that unusual. My FX56 used to IDLE @ 38+C, 1.42V. Temps would easily skyrocket into the 60's when loaded. The opties seem to be hotter cpu's in general. I'm also thinking that the thermal paste has something to do with the high temps. Ever notice that on almost every pic of a cpu and it's lid taken off that there is NO paste on the core and basically all of it still stuck to the IHS? ;)Quote:
Originally Posted by bachus_anonym
EDIT: I don't feel like it's a hijack if it's informative and interesting....
I may try sp2004 since I just remembered that the multi-proc tweak may produce a little TOO much heat for comfort. That was one of the issues with my win2k3 install. For some reason, my cpu would start loading up for no apparant reason (no, the multiproc hal wasn't installed). I'd be doing nothing in windows and my temps would be @ -30C when they should have been in the -47C range. That's why I've done all these spi runs on winxp....:p:Quote:
Originally Posted by Vapor
Nope, a divider won't really help. It's the HTT which bothers it. But in the past, I've been somewhat successful by dropping the multi by .5. For some reason, that allowed some cpu's to be able to run a higher HTT and clockspeed with a little more stability...:confused:
http://img489.imageshack.us/img489/1950/a9jy.jpg
You got a bit of tweaking to do stephen, seems a bit slow for 3.8 ;)
DO NOT do multi cpu tweak. Bad for pi times. As for temps with my 146 I saw 240c idle with multi tweaked win 2k vs 220c in untweaked xp. Try a bit more vcore.
I use RightMark CPU Clock/Power Utility in the battle against clock bug.First run it and go to advanced option it looks like this:
http://www.xtremesystems.org/forums/...1&d=1139776848
Then go to ACPI State to Show/Modify and select C1 (HALT):
http://www.xtremesystems.org/forums/...1&d=1139776848
unckeck the CPU Low Power Enable and click Apply:
http://www.xtremesystems.org/forums/...1&d=1139776848
Idle temp now never goes below -20c and stays around -15c on a Hipro5 modded Prometeia Mach I with R507 and i can increase HTT with nothing running in the background.Mine 144 over 340HTT freezes in windows at -26c and ACPI Multiprocessor never helped it..........
There's a few things that make mine and Jason's? run different:Quote:
Originally Posted by afireinside
1. Look at my Vcore and look at his. Vcore can make a huge difference in terms of spi results. He's using .1V more than me and I've seen that translate to almost .1 Sec difference in times. So can Vdimm....
2. He's using Win2k3. I'm stuck using winxp. There's an added stability advantage to using win2k3 and as we all know, stability = speed in spi.
3. That Spi time seems a little fast to me @ that speed. It may be possible, but I'd be curious to see his tweaker settings and voltages. Also, it would be nice to know if he was using maxmem or a ramdrive. I was using none of those......
I'm not sure that result can be duplicated. I'm curious to see if anybody else has been able to run that time @ that speed, although I doubt there are many out there in the first place......
I'll see if I can improve on the score later.....
Thanks for the suggestion. Looks like an interesting prog. Got a link? And just curious how much does it raise cpu temps?Quote:
Originally Posted by ThorLite
http://cpu.rightmark.org/products/rmclock.shtmlQuote:
Originally Posted by s7e9h3n
On mine SS goes up 10c when its idle only,load temps dont change and i can increase HTT up to 371MHz on the 144 with out it i cant go over 340MHz or i must load the cpu with something not to freeze.........
http://valid.x86-secret.com/show_oc.php?id=67214
SOunds awesome if it's capable of what you claim. Thanks again :toast:Quote:
Originally Posted by ThorLite
Oh yeah - and load temps shouldn't ever change regardless of how many progs are running. Only thing that would affect load temps is Vcore ;)
Direct download link of the proggy:Quote:
Originally Posted by s7e9h3n
http://cpu.rightmark.org/download/rmclock_20_bin.rar
I hope to make good use of it so you can get even better scores from your wonderfull opteron.... :D :D
Really amazing!!! :woot: :woot: :woot:
But i'm not sure if i understood it in the right way, but you (stephen) said that if you run a spi @3ghz 1.4v and then @3ghx 1.5v the second will be better? :)
thank you and :clap: :clap: :clap:
Zac
Nice prog.
Steve, already tried it? :)
Sorry for the confusion...let me explain. Spi is a very sensitive bench as there are many factors which comprise a great Spi run. One such factor is Vcore on the cpu. If you have barely enough vcore to complete the run at the given speed, then you'll finish with a slower time than in a run with sufficient Vcore. Try it out - run a spi with the minimum Vcore it takes to finish the run, then give it a little bump. You'll see the difference in times. Also, it's possible to hurt your times if you run TOO much Vcore for that speed. Hope this helps....Quote:
Originally Posted by Zac89
Lol...nope not yet...I'll try later tonight...I'm out the door to go pick up my GPU Phase Change cooler and won't be back till tonight :cool:Quote:
Originally Posted by PoL
ok ok now it's clear I think... so the thing that matters is the stability of the cpu, right? if it gives some rare errors like when you are near the limit, spi will slow down... tnk you :)
:up: Remember - stability is king in Spi. It's not a matter of tight latencies on the memory or enough vcore to kill an elephant - it's about getting the benchmark stable at a given speed....;)Quote:
Originally Posted by Zac89
Thanks for that little bit on spi!
Amazing results! :toast: :toast: :banana: :woot: :clap:
Nice results man! thats gotta be the best opteron made... to bad about the bug hopefully you can find a way around it. Haven't you said before that DICE is a bit "warmer" than ur autocascade? could give that a try, or a really good SS running R410a or something similar. She's got more left in the tank, good luck squeezing every bit of it out of her!
Congrats on that nice Chip ! My Venice also stables @ 2.6 with 1.5 but even if i give it 1.7 won't stable at 2.7, bench's at 2.9, screenie is about 3.1... Weird Chips...
Not sure I agree about load temps, run prime for the same length of time you run spi 32m and you should see a difference.. maybe you don't notice on your auto, but not all 100% loads are created equal... or try toast or some of the other heat generating programs. Personally my pi runs put me anywhere for 3-5C cooler than any prime or toast even at the initial load temperature spike.Quote:
Originally Posted by s7e9h3n
I'm talking about ABSOLUTE load temps....say if you ran Toast as the only program running. If you were to run prime alongside it, your temps would top out at the same place......100% load is 100% load. There's no such thing as 95% 100% load :p:Quote:
Originally Posted by Weee
I have a feeling this cpu would kick the crap out of ANY single stage short of a 1hp compressor lol...just ask Dumo...his EVAP temps went positive on his 154 and a vapols :eek:Quote:
Originally Posted by Fhqwhgads6680
not sure I agree with this...Quote:
Originally Posted by s7e9h3n
all depends on what part of the CPU is really loaded. FPU load is much different that memory access load for an AMD CPU...but a non-blocking 100% load is still read the same even though some programs definately show higher temps than others... :slap:
Because it's a vapoLS AKA a piece of junk. But if his EVAP temps are going positive, theres something wrong. A well build single stage should hold AT LEAST -30c if not -40c on the evap with a 154.Quote:
Originally Posted by s7e9h3n
Both 3dmark and SP2004 take me to 100% load but SP2004 is a few C hotter.
100% is 100% period. If it's using part of the cpu and not another, then it's not truly 100% loadQuote:
Originally Posted by freecableguy
Try a 154 @ 1.71+ volts and then see if you can make that claim. It's simply the hottest running cpu I've ever seen and -as I said earlier- an x2 @ 1.65+ loaded doesn't even compare as I've tested both.......Quote:
Originally Posted by afireinside
Steven,
I think AFI is right... It's clearly impossible to see positive temps on an evap provided cooler is working properly... Positive in MBM, yeah... In BIOS - maaybe...
Before that 146, I had a chance to play with Eric's Opteron 150. It was THE hottest thing I ever, ever played with! SmallFFT would load it to 15C in MBM @ 3.2Ghz 1.62v!!! :shocked:. BIOS temps at the same time were -12C!!! I re-mounted the head like 3 times because I could not believe my eyes :D Yet, evap stayed at -39C LOAD (during that SmallFFT "challenge"). This is when you know, that IHS is a major limiting factor. But not all Opterons are like that and I think that it's just a far generalization that Opterons are somehow hotter... Remove the IHS off of any of those "hot chips" and you'll see temps drop drastically.
My 146 ("very warm" with IHS) without IHS is like 10C cooler than my 3700+ was (with IHS but it was a "cool/cold")...
Yeah I think I am noticing this on my 148. I'm at 3.255 ghz at 1.71 vcore, and on my single stage the cpu (read by speedfan) idles at -25 but under full load (I.E. prime or spi) it drops to about -5....seems to be a pretty large delta between load and idle even for a single stage...Thinking its the IHS...but I've never poped the top on a cpu and if I eff it up...then im out a computer for a while.... :(
How were you reading bios temps @ load? I'm telling you - push 1.71V into that cpu and see how it handles on your single stage. I promise you - there's a WORLD of difference between 1.62V and 1.71-1.76V. It also matters where the probe is mounted on your evap. If it's mounted on the side of the evap, that may tell you nothing in terms of what the temps are like on the FACE of the evap. Removing the IHS off this cpu didn't help one bit as the heat is too concentrated in one area and actually overloads the evap in that spot.....you can't compare the 146 or 148's to these cpu's. I'm pretty sure I recall seeing someone here on the forums with a 150 which IDLED @ 50C on air....Quote:
Originally Posted by bachus_anonym
All 0522 TMPWs are hotter than any other FX or Opty (makes no difference) that I have laid my hands on.
Including your FX60?Quote:
Originally Posted by mikeguava
OT: Mike - give me a call if you're still awake :p:
I mean, it's -12C in BIOS @ 3.2-3.3GHz 1.63v. With this 146 I have here it used to be -16C (BIOS, IHS On) and now -22C (BIOS, IHS Off, 1.72v, 3.6GHz). My SD 3700+ sat @ -19C (BIOS, IHS ON, 1.80v, 3.4GHz).Quote:
Originally Posted by s7e9h3n
I will be glad to test that for ya the moment I find that 154 in the mail :D :p:Quote:
I'm telling you - push 1.71V into that cpu and see how it handles on your single stage.
I remember Reggie posting a diagram showing where exactly the probe is placed in my cooler's case. As far as I remember it's about 5-10mm above the face of the evap, as close to the middle as it was possible... I'm gonna ask him again, though.Quote:
It also matters where the probe is mounted on your evap. If it's mounted on the side of the evap, that may tell you nothing in terms of what the temps are like on the FACE of the evap.
But I thought you said and showed us that removing IHS helped you? :confused:
What I'm trying to say that I assume/hope that with all those Opterons that people see are very hot, quick fix of popping the IHS will make a world of difference for them - at least in terms of temps. I wish I had enough cash to keep that 150 from Eric so I could remove the IHS and prove my little theory to you :D Because I'm 100% sure that his 150 was even hotter than your 154 ;)
No, in fact , removing the IHS made things WORSE. Read a few posts back about me needing a "cold-plate" to distribute the heat better. The naked core was basically overloading my evap. My temps were ~15-20C WORSE without a medium between the core and the face of the evap. Reggie mounts his probes on the side of the evaporators ~1/4 of the way up from the face. For me, it's tough to tell temps using that probe, so I always insert a probe into the socket area touching the ihs of the cpu (obviously when it's still on). OK, let's do the comparison thing then :p: :Quote:
Originally Posted by bachus_anonym
FX57 @ 1.74V - Bios temps @ -30 - -35C
Venice 3200+ @ 1.94V - Bios temps @ -25 - -28C
Opty 154 @ 1.71V - Bios temps @ -18 - -25C
There's no way in hell that could be true. My LOADED temps on a FX57 @ 1.71 - 1.74V is read my MBM @ -30C. The day your single stage could do that is the day I believe you. You have NO IDEA what kind of capacity this autocascade has and until you do, I'm sure you won't believe me.....;)Quote:
Originally Posted by bachus_anonym
Edit: Now that I think about it, I'm wrong about the -30C...it's more like -28C..sorry....
Haha... I think there is no point to this argument :slap: The only way to find out if your 154 is exceptionally hotter than anything else around, is to put it to test on something else than your auto-cascade :)
Let me know when you get bored with this 154 so we could do a little swap, just for fun of testing - your 154 for my 146 or 3700+ :D Isn't that a hell of a deal?
In addition to my screenies from post #144, I just run 15mins SmallFFT w/ 146 @ 3500MHz, 1.76v CPU-Z (1.78v DMM)
MBM @ -6C and stays between -7C and -5C
EVAP @ -40.5C LOAD
BIOS (quick RESET by pushing RESET button, SP2004 not stopped, 7s from SP2004 LOAD) @ -14C
-----
BIOS (after stabilizing) @ -18C
BIOS EVAP @ -44C
Now I wonder what temps I would be getting with that 154 of yours, in exactly same conditions... But just for kicks, can you run it on your cascade in a same way and tell me all the data I gave you, please? Thanks :toast:
Bachus I think that 154 is ALOT hotter than my 150 was but you never know. You would prolly see like +20:)
I'm not saying it can't be... It might be but like you said - you never know. What I'm saying is that there is just no way that I this chip will bring single-stage down on it's knees and let it cry over positive evap temps. But again, there is no other way to see than actually test it on a single stage... :) I have no problem being a guinea pig for that test, though :D Just say a word, Steven ;) :toast:Quote:
Originally Posted by njkid32
Also, is there someone here than can explain why would Opteron 154 be hotter than e.g. 150 @ 2800MHz, no IHS (so we can eliminate crappy TIM or bad contact) and under exactly same cooling conditions. I just don't believe any physical/design differences between cores of 146 OCed @ 2.8GHz and 154. Anyone would like to try to explain that? :)
It's not a difference in design - it's a difference in silicon. ;) I do have a single stage I can test on eventually - it's a mach2. But that's at Reggie's house. I also can try out my Baker Gpu cooler on the cpu for kicks, but that's a 1hp Rotary Single stage which is most likely comparable to some of the strongest single stagers every built. Here's what the evap does on that unit:Quote:
Originally Posted by bachus_anonym
http://www.cryo-laboratory.com/uploa.../Baker/-70.jpg
But just think about it for a sec...Quote:
Originally Posted by s7e9h3n
If there's no difference in temps between SD 3700+ @ 2.8GHz and FX-57 @ 2.8GHz (which are identical cores, 1MB L2, etc) then why would 154 @ 2.8GHz be hottter than 3700+ @ 2.8GHz or 146 @ 2.8GHz??? We're talking "normal" chips, not some hot-a$$ samples ;) If there's anyone that can explain that from the technical point of view and info he provides cannot be questioned then I will give up to the claim that Opterons 154 are way hotter than Opteron 146-152 OCed to 2.8GHz :)
IDK why either...but I gotta say, clock for clock, volt for volt, my 146 was just as hot as my 4400+....these Opties get hot for sure. As for why the 154 is even hotter than the rest of the Opties....I imagine there's some internal differences for additional stability. A pin-to-pin resistance comparo *might* lead to some clues, albeit tedious.
How a 'San Diego' core could dump out 175W+ is interesting, almost like the true San Diegos are tweaked down to acheive stability through lower temperature stress while the top-end Opties are tweaked for stability via just keeping all the electron flow in check and keeping 'resistance' high, knowing that the heat will be handled.
That's just my take on *why*, I flatly don't know the answer for the *how* they're so much hotter. :(
EDIT: was I speaking English before? jeeze, that made no sense
I'm not sure just what you're trying to say? Clockspeed is not the major factor which dictates core temps. It's the silicon and vcore :confused: BTW, where did you get the info that a SD 3700+ @ 2.8 is the same temp as an FX57 @ 2.8? Unless you can do 2.8 @ default vcore with the 3700+, that would be quite difficult to imagine....Quote:
Originally Posted by bachus_anonym
I was talking about 2.8Ghz and same voltage in all those instances, say 1.35-1.40v...Quote:
Originally Posted by s7e9h3n
But I think there's no point in further discussion. None of us can really proof his point because I don't think we have enough technical knowledge (at least I don't) that would explain potential 10-15C differences between chips. If one might think that better silicon gives better/cooler chip then why 154 (which should be superior quality silicon) would be so hot?
But like I said, we would have to drag an AMD engineer here and give us a lecture :D
I forgot to mention one thing though. I've had a talk with my AMD guy about something similar to this topic in the past. I had mentioned the fact that FX cpu's seem to run a little hotter than the normal San Diego core cpu's. What he told me was quite interesting: FX's share a similar memory controller architecture as the Opterons. This is evidenced by examining where the FX line of cpu's evolved from. Remember what the first FX cpu was? It was a s940 FX51. Yup, it was basically an OPTERON with an unlocked multiplier. As interest grew in these FX's, AMD decided to move the cpu to the 939 platform where consumers would more easily have access to this "enthusiast" cpu. So the s939 FX's were born - San Diego cored cpu's with an Opteron's memory controller. He was saying something about (as Vapor mentioned) how the Opties tend to run a little hotter than the regular A64's due to their memory controllers. I don't recall him saying why they were hotter, just that they were. I'll give him a call tommorow and see what I can dig up, but I'm almost positive that a cpu's heat output is VERY much dictated by the silicon which it's built upon.......and in this case it seems as if the 0522's were exceptionally warm cpu's (and MikeGuava seems to agree)......Quote:
Originally Posted by bachus_anonym
That 150 is a hot mofo. My 148 is almost that hot, priming at like 3.35 with 1.63 volts, got to +14c.Quote:
Originally Posted by bachus_anonym
I put the 148 on air with my ACFreezer64 Pro, 1.6 volts, started Sp2004, temps shot INSTANTLY to 60C, it scared the crap out of me.
All i can say from my Opteron testing (i've been a venice guy) is that they run SO hot. I know my venices didn't load at 52c on stock volts/stock cooling.
thats a nice opty u have there ;)
It Benches in 3D as well :p: This is just a "warm-up" run - untweaked and unoptimized ;):
http://service.futuremark.com/compare?2k3=4621125
http://www.cryo-laboratory.com/uploa...850/03Rank.jpg
Damn Steve, thats really flippin nice! :D
What were the cardclocks for that run, and voltages?
Ummmm....you don't see them :p: ? 789/690 @ -59C. 1.72Vgpu, 2.4VDD+VDDQ. BTW...I owe you a run....I'l try to get it done later ;)Quote:
Originally Posted by JasonDTM
Already putting the GPU cooler to good use. Very nice. And a belated congrats on 4GHz. :toast:
Verrrry nice Stephen!!
Thanks my friend - a big thanks to you for the Baker unit! :toast: It's kinda funny - my probe on the evap read -51 when I booted, but ATI tool was reading -59C on the card :wth: There definately more left in the card and the cpu.....Quote:
Originally Posted by sierra_bound
^^^ lol I'm blind as a bat, and with my clocks in 01 and higher mem speed I think 33k to 34k is probably feasable.
PM me again as to what exactly you wanted me to run....I'll do that when I attempt to get an 01 score maybe today or tomorrow.....Quote:
Originally Posted by JasonDTM
Nice 03 score with the X850 XTPE. Good to see you doing some 3D Markin'. We know that you have the CPU's to do it ;)
Good luck pushing some more in 03 and attempting to get some 01 scores.
Oh btw, Im coming back into overclocking etc. Gunner get my self a good strong single phaser. I want the 3800+ Venice back :p What a brilliant cpu you sold me :D I need to try to buy it back ;) Do some 'non' Opty overclocking... hehehe...
Awesome scores s7e9h3n !!! :eek: Congrats from Poland :toast:
And, of course, keep OCing ;)
Last night I was about to ask you If anything more you were cooking with this 154 :)Quote:
Originally Posted by s7e9h3n
Very nice 1st shot at 3DM01 with that Opty :up:
Lol...that was 3d03 ;) This is 3d01...we'll see how long this stays at the top of the Orb before k|ngp|n posts his backups :p: This was a test run anyways - no tweaks - just a straight run-through:Quote:
Originally Posted by bachus_anonym
http://www.cryo-laboratory.com/uploa...n/x850/45k.jpg