Wow--New Black Edition Phenom Next Week-- Comments?

Page 5 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.

epsilon84

Distinguished
Oct 24, 2006
1,689
0
19,780


Is that supposed to be an impressive overclock on a Phenom? Seems a rather high voltage to get 3GHz, but then again what do I know, I've only been overclocking C2Ds...
 

bfellow

Distinguished
Dec 22, 2006
779
0
18,980

I know this doctor named Brown who is almost done working on a flux capacitor.
 

Pepethebandit

Distinguished
Dec 18, 2007
6
0
18,510
Wasn't the TLB Errata only suppose to affect Phenoms above 2.4ghz in a select benchmark?(Locking up I believe) I have to go back and research when the errata first came to light but that would explain how you can have phenoms clocked higher then 2.4. Found an Inquirer link http://www.theinquirer.net/gb/inquirer/news/2007/11/18/amd-delays-phenom-ghz-due-tlb about the TLB Errata. I can put up with dealing with the errata if what the inquirer says is true.

Unlocked Phenoms...mmm..that might just make me finally upgrade. Since I plan on gaming at 1400x900 for a while to come I don't need xtreme anything.



Current Rig: AMD 3800x2@ 2.4(939)-a8r32-mvp 3 gigs-x1900xtx-80gighd-500gighd-audigy2-samsung 19inch wide
 

jevon

Distinguished
Jun 6, 2004
416
0
18,790
But I think what he's saying is that it's almost impossible for an end user to reproduce the error, ie it basically can only be done in a lab setting. Anandtech and the enthusiasts haven't been able to cause the crash, so I'm sure most average joe overclockers and such will be fine :) I thought I had read on the NCIX boards that the QX9850 had a FSB error in it that's basically the same idea, virtually no home user would ever trigger it?
 

spoonboy

Distinguished
Oct 31, 2007
1,053
0
19,280


Agreed, they know the hows and the why already.
 

spoonboy

Distinguished
Oct 31, 2007
1,053
0
19,280


As far as overclocking goes, prime stable is stable.
 

husky mctarflash

Distinguished
Dec 3, 2007
215
0
18,680




Wow. This thread's come quite a long way. Some moments of sadness, some of laughter, and some of...well...utter ridiculousness.

We now see one more "confirming" report of immanent Phenom BEs, but is this all legit? Will we really see such a release this week? Ever? Or is it smoke and mirrors? Testing the water?
 


Running a CPU-intensive program like Prime95 for many hours without errors is a general predictor of overall stability. However, it is not the end-all, be-all and I'd suggest running other programs that are stressful but sensitive to errors like Orthos and Folding@Home or even games (to check for locking up and gross instability, not the occasional hiccup that does not cause a program to segfault.) FAH is probably the most sensitive program I have run across and I recommend its usage, especially if you complete a work unit or two as that takes several days.

Oh, one more thing: watch those temps. "If it's too hot, it's not stable."
 

homerdog

Distinguished
Apr 16, 2007
1,700
0
19,780
Agreed, I've had completely prime stable systems (8+ hours on both cores) that locked up during everyday browsing. It was my memory, and for some reason Prime wasn't effected by it.
 

ro3dog

Distinguished
Mar 20, 2006
243
0
18,680
Kyle Bennette at hardocp has his oc to 3Gigs with no errata patch along with other folks that bought the Phenom.The question I ask was statement,so I pose it as a question here ,which so far I have no answers.Everyone talks about the bug but nobody has shown it,reviewers have shown the penalties of the errata patch but thats it
 

azfj60

Distinguished
May 3, 2007
46
0
18,530


I can kind of see where you're going with this, but stop and ask yourself this... Would AMD put the (punishing) errata patch in place if it wasn't absolutely necessary? Errata that tank every processors off the production line, all the time, in every computer in which they are installed, are usually caught in the modelling or simulation phase of CPU design/manufacture. Be content that this is likely a very rare bug, but one which AMD realized needed a fix in place. Add on to that the "PR factor" which Intel learned the hard way...

(Likelihood of bug)*(impact of occurrance)*(negative press impact of bug) = probability of implementing a "fix"...

 


If "The Bug" does not exist, as you suggest, then why has AMD released a patch for it? AMD also acknowledged its existence during it analysts meeting.


"Phenom TLB patch benchmarked" - http://techreport.com/articles.x/13741

"AMD Phenom is BROKEN - Costs a 14% performance hit to fix" - http://www.hardcoreware.net/amd-phenom-is-broken-costs-a-14-performance-hit-to-fix/

 

BaronMatrix

Splendid
Dec 14, 2005
6,655
0
25,790



Wow, that's perhaps the most negative report yet. Congratulations. It should say BIOS workaround costs 14% perf.
 

homerdog

Distinguished
Apr 16, 2007
1,700
0
19,780

I don't doubt that the occasional Phenom sample can hit 3GHz, but the majority will not without extreme voltage and cooling.

I also assure you that the TLB errata is very real, and that the "patch" to fix it is not optional and will decrease the per clock performance of Phenom by over 10%. AMD has stated that they will release an overclocking utility that will allow you to disable the patch, but the details on this are vague.

Edit: Jeeze, I just realized that the Northbridge on retail Phenoms is only 1.8GHz, 200MHz shy of the engineering samples that were initially reviewed. AMD, this is not acceptable and you will not gain anyone's trust by doing things like this.
 

coret

Distinguished
May 29, 2007
273
0
18,780
Yes the errata is real ... but have any review sites been able to reproduce the error? Yes or no?

Can we have a straight answer on that point, instead of it being deflected by "well it's there or else why is there a patch" comments.

If not, then there's no point in the majority of people applying the bios fix.
 

cnumartyr

Distinguished
Nov 3, 2007
2,287
0
19,780



Odds are it's been blown way out of proportion.

Talking to a few people that have them they don't have any problems with the TLB bug. They do have serious issues with heat, overclocking and voltage. I really think the TLB should almost be non-issue as all processors have errata... AMD just handled this one poorly and tried to hide it... Most of the guys I've talked to are topping out around 2.6-2.8 GHz. I'd like to see how B3 can correct this or if it's just a TLB fix.
 

ro3dog

Distinguished
Mar 20, 2006
243
0
18,680
I'm not saying there's no bug,because AMD said there is,but which reviewer has stated that they have seen results of the errata.I think AMD really shoot themselves in the foot with this TLB thing ,for there are alot of folks that are really enjoying there Phenom after testing the waters.From what I have see,the present Phenoms are clearly no match for what Intel has,however they are far from the pictures that have been painted about it.The B.E Phenoms will be out next week with new reviews.
 

homerdog

Distinguished
Apr 16, 2007
1,700
0
19,780
The point is that it doesn't matter how rare the errata is. It doesn't matter whether or not you are likely to ever see it in the life of your processor. The fix is applied behind the scenes in the BIOS and you will lose performance regardless unless you never update your BIOS.
 

cnumartyr

Distinguished
Nov 3, 2007
2,287
0
19,780


Agreed.

Patch sucks and should be optional. :D