Have you experienced TLB errors with your Phenom?

Have you experienced TLB erros with your Phenom?

  • Yes

    Votes: 4 13.8%

  • No

    Votes: 25 86.2%
  • Total voters
    29
C

ClariorHincHonos

[H]ard|Gawd
Joined
Jun 25, 2007
Messages
1,972
Pretty simple, yes or no? If you have, please post details.

FYI...this is just an information gathering thread. This is not to bash or promote AMD. Let's keep it neutral and straight to the subject-whether or not you have experienced TLB errors. THAT'S IT.

Thanks!
 
I think this thread may have called for a visible poll :p
*subscribed for interest*
 
FYI: The TLB error will happen under very specific conditions - if you have experienced these errors, please take a moment to describe what exactly you have been doing and what exactly has happened. A few of my engineers would love to read about it.
 
Hell no,and I have built and used three Phenom systems.I did expireince some instability though,and this was at stock clocks,and overclocked.

Using Crucial Tracer ram,PC800 and 1066 DDR2,and a MSI 790FX board as well as the Gigabyte 790FX motherboard.I have a Asus M3A32-MVP Deluxe WiFi sitting here waiting for a BE Phenom.When I have that system built up,I plan to run all manner of software,and
see how far I can oc it.

The crashes and freezes I had happen with all three builds,were for the most part when I was trying to oc them.Only the MSI board consistently crashed at stock speeds,while on the windows desktop or in the bios.Sometimes even just looking around in the bios on the MSI board would cause crashes. :( I had none of these issues arise with the Asus board.
 
Again...to those that vote Yes-describe your situation. Otherwise I'm just going to view those votes as erroneous/malicious. I know that sounds kind of a-holeish, but if you describe your results it may help correct the problem, and help others avoid the issue. Thanks in advance!
 
nobody is going to explain anything (for the record I did not vote)
You asked for Intel trolls to rape this poll by making it anonymous.

Plus I dont see the point of this....of course the vast majority of people won't have the bug.
doesnt change the fact that it does happen. So what are you trying to prove?
 
Yeah if they've run into it I'd be curious to know what caused it.

I heard that VMWare could/may/will run into the errata, but since virtually no one has a Phenom and even fewer use VMware it is difficult to get any solid info,
 
osalcido said:
nobody is going to explain anything (for the record I did not vote)
You asked for Intel trolls to rape this poll by making it anonymous.

Plus I dont see the point of this....of course the vast majority of people won't have the bug.
doesnt change the fact that it does happen. So what are you trying to prove?
Click to expand...
It was asked for by someone else in another thread, the purpose being to put the whole TLB deal to rest. I'd also like to know if it does manifest in a normal desktop/gaming environment, because Phenom is an upgrade option for me.

Thanks for the "advice" about making it anonymous. I'll keep that in mind next time.
 
NKDietrich said:
Yeah if they've run into it I'd be curious to know what caused it.

I heard that VMWare could/may/will run into the errata, but since virtually no one has a Phenom and even fewer use VMware it is difficult to get any solid info,
Click to expand...

More and more people everyday are sporting one, brother! ;)
 
NKDietrich said:
Yeah if they've run into it I'd be curious to know what caused it.

I heard that VMWare could/may/will run into the errata, but since virtually no one has a Phenom and even fewer use VMware it is difficult to get any solid info,
Click to expand...

I think both of those assumptions are bad....

I have read several articles (need to look up links to post here, actually) that Dell will _begin_ selling boxes with dual OS's running using vmware. When that becomes standard and people load down their box I could imagine it, maybe, being a bit more common to see the TLB bug. I suppose Dell would have to sell Phenom based boxes first, though :)

Any box I have to do any serious work on has vmware installed, or there is a vmware server on the lan that I have a console into. Pretty much everyone in our engineering group is the same.
 
Is there something that can test for the TLB error, like the specific sum that broke the original Pentium FPU, and if not how would you know you've got the bug? Surely not just a crash, that could be anything.
 
Manny Calavera said:
Only the MSI board consistently crashed at stock speeds,while on the windows desktop or in the bios.Sometimes even just looking around in the bios on the MSI board would cause crashes. :( I had none of these issues arise with the Asus board.
Click to expand...

Did you try with the *same* CPU on the Asus board and MSI board?
Else it could be that the particular CPU was more sensitive to the bug, and it wasn't the motherboard that was unstable.
 
JDAdams said:
Is there something that can test for the TLB error, like the specific sum that broke the original Pentium FPU, and if not how would you know you've got the bug? Surely not just a crash, that could be anything.
Click to expand...

Could it? Modern systems don't crash on me anymore. Some software may hang occasionally, but Windows will catch the exception, kill the task, and you're back on. It's always the same software that crashes aswell.
So as soon as I'd start to see random crashes... software crashing that never crashed on other systems... or the entire OS hanging... I'd be suspecting the CPU quite quickly to be honest (especially when I would be using chipsets and videocards etc that I have also used in combination with other CPUs and didn't have problems with).

It could be anything, you never know for sure... but it would be very suspicious if a Phenom crashed in all kinds of situations where other systems never crash.
 
Scali! You're back!

On topic: If you want to test for the TLB issue, AMD has outright admitted that the error happens when virtualization is in play. Thus, if you wanted to go hunting for it, grab a free copy of VMWare Server or Virtualbox and use the heck out of a virtual machine.
 
I'm aware of that, but if it then crashes, how do I know it was the TLB problem? I suppose if you could consistently reproduce it with a certain set of instructions that would do it, but from what I've read on the error it's just a race condition so will arise pretty randomly. Unless I was running a production server I'd be inclined to put up with the odd crash, any decent machine should be able to recover anyway, assuming it does crash and doesn't just hang. We have Linux servers at work with threads that randomly hang on occasion, never has this been attributed to a CPU bug, it just happens (that's what our 3rd line suport say anyway). As for Windows I downright expect the odd glitch now and then whatever the platform is.
 
JDAdams said:
I'm aware of that, but if it then crashes, how do I know it was the TLB problem?
Click to expand...

Let's face it: us regular users will never be able to know for sure.
AMD and OEMs have special test beds with all sorts of equipment attached to the motherboard to analyze the inner workings in detail, and reproduce certain situations.
We can only suspect that random crashes come from the TLB bug. For all we know, it could just be a faulty motherboard aswell, or something silly like that.
In the HardOCP labs they may be able to switch components around and do a bit more structured research as to why certain systems crash sometimes, but even that is mostly a gamble. As us overclockers should know... when you're on the edge, you may have a system that's stable for a few hours at a time, but it might just aswell hang a few seconds after it's powered on. It's not just one thing that triggers the crash... It's a combination of many factors.

Those who know a bit about electronics will probably know that a transistor doesn't have a fixed propagation time. Rather it has a minimum and maximum propagation time, and the actual propagation will be somewhere in between, but within that range it's pretty random.
As far as I understood the problem, what happens here is that the TLB *sometimes* doesn't update certain flags on the table entries quickly enough (that's what the linux and BIOS workarounds seem to do, avoid the usage of some flags)... in other words, the maximum propagation time is longer than the system expects. Most of the time it's fast enough though, so you 'get away with it'.
 
Scali2 said:
Did you try with the *same* CPU on the Asus board and MSI board?
Else it could be that the particular CPU was more sensitive to the bug, and it wasn't the motherboard that was unstable.
Click to expand...



No,each system I built used a different Phenom chip,two 9500's and a ES 9600.The ES would go as high as 2.7~ oc'd,but was only OS stable,not 3d,etc.


Edit : Welcome back Scali2 from your 'vacation' :D I guess that means Duby may also be coming back as well ...
 
ive had a 9500 in my pc for a week with the tlb patch disabled and havent had any crashes yet..

that said i dont run any of the sort of software that might cause the bug afiak
 
You must log in or register to reply here.
Back
Top