Sign in with
Sign up | Sign in
Your question

Users of Phenom in standard AM2 platform

Last response: in CPUs
Share
January 6, 2008 5:08:59 PM

Hello all.

I would like to hear comments from users of the Phenom Chip in their standard AM2 platforms.

I currently have an ASUS Crosshair mobo. Per my research it's a direct drop-in with the proper BIOS flash.
Was there a performance gain of any type over your dual-core chips?

Seems like there are mixed opinions and a lot of disinformation I am having to wade through.
What about REAL educated opinions from REAL users?

I am running a 5000+BE at 3.1 ghz with 2 gig of RAM right now and I am pretty happy with it. I was planning on waiting for the new stepping, but was wondering what the observations were so far with the current steppings.
I hear some people actually think they are pretty good.

There's gotta be some benefit to 4 cores over the two I have, even with the standard hypertransport of the AM2 platform no?

Let's hear some opinions from current users.

Please do not make this an Intel vs AMD thread.
Thanks in advance.

January 6, 2008 6:02:04 PM

I would not trade in your 5000+BE for a Phenom. I myself don't have Phenom, although I have flashed my motherboard to be compatible.

You have your chip running at an excellent speed; with Phenom you can only dream of clocking that high. That being said, yes 4 cores will benefit you in multithreaded apps so if you're into heavy encoding/video editing then I would suggest you wait till the B3 stepping of Phenom before upgrading.

Otherwise, and this includes if you are a gamer, I would stick with your current processor - the high clock speed will mean that it outperforms the Phenom in single/dual threaded apps.
January 6, 2008 6:14:20 PM

I'm just waiting for the non-Phenom owners to come in here shouting "ZOMFG TLB ERRARTA PHENOM FTL!!!!"

I give it 15 minutes...
Related resources
January 6, 2008 6:15:48 PM

I was thinking perhaps to dig this thread up with 2nd Q comes around. :lol: 
January 6, 2008 6:20:26 PM

There are several Phenom owners talking about their processors in this thread at xtremesystems

You will have to read the whole thread, but its only 2 pages
January 6, 2008 6:26:35 PM

This is worthy of a God-Like post...

The wisdom of the thread ... truth of the matter, systems lock up for various reasons, trying to attribute to this specific bug would be difficult at best... as the other platform bugs get squashed, the rate of occurrence of any lockup would most likely not be anything noticable or outside the norm...

[b said:
Most people would simply curse M$ anyway.... heck, the last round of 'Vista updates' wreaked havoc on one system, I had to go back with system restore 5 days just to get back to where I am day on day stable again.]The wisdom of the thread ... truth of the matter, systems lock up for various reasons, trying to attribute to this specific bug would be difficult at best... as the other platform bugs get squashed, the rate of occurrence of any lockup would most likely not be anything noticable or outside the norm...

Most people would simply curse M$ anyway.... heck, the last round of 'Vista updates' wreaked havoc on one system, I had to go back with system restore 5 days just to get back to where I am day on day stable again.
[/b]

This was posted by JumpingJack at the link Turpit provided.


So what he's saying is you wouldn't even consider a crash to be form the L3 cache, you'd probably suspect Windows and just simply hit the reset button on your PC. TLB bug is over-rated at best.

From the users posting there it would appear as though Phenom is just fine. Imagine that.


But a strong point needs to be made.... that window of time is likely very small, so small that the probability of happening is slim to none ... [b said:
this errata does not give cycles of time that it takes to update the TLB and even if it is 10's of cycles, it is such a small window, for typical DT usage one may never actually trigger it and if it does, the frequency of M$ blunders would overwhelm the signal.]But a strong point needs to be made.... that window of time is likely very small, so small that the probability of happening is slim to none ... this errata does not give cycles of time that it takes to update the TLB and even if it is 10's of cycles, it is such a small window, for typical DT usage one may never actually trigger it and if it does, the frequency of M$ blunders would overwhelm the signal.
[/b]

Also posted by JumpingJack. The bug is so rare and the window of opportunity for it happening is so small you would probably NEVER encounter it.


Also, explanation of the Errata bug...

The processor operation to change the accessed or dirty bits of a page translation table entry in the L2 from 0b to 1b may not be atomic. A small window of time exists where other cached operations may cause the stale page translation table entry to be installed in the L3 before the modified copy is returned to the L2. In addition, if a probe for this cache line occurs during this window of time, the processor may not set the accessed or dirty bit and may corrupt data for an unrelated cached operation. The system may experience a machine check event reporting an L3 protocol error has occurred. In this case, the MC4 status register (MSR 0000_0410) will be equal to B2000000_000B0C0F or BA000000_000B0C0F. The MC4 address register (MSR 0000_0412) will be equal to 26h said:
The processor operation to change the accessed or dirty bits of a page translation table entry in the L2 from 0b to 1b may not be atomic. A small window of time exists where other cached operations may cause the stale page translation table entry to be installed in the L3 before the modified copy is returned to the L2. In addition, if a probe for this cache line occurs during this window of time, the processor may not set the accessed or dirty bit and may corrupt data for an unrelated cached operation. The system may experience a machine check event reporting an L3 protocol error has occurred. In this case, the MC4 status register (MSR 0000_0410) will be equal to B2000000_000B0C0F or BA000000_000B0C0F. The MC4 address register (MSR 0000_0412) will be equal to 26h


Seems as though the only people making a fuss about this is the fanboys and just over-all dumbasses.

January 6, 2008 6:30:25 PM

justinmcg67 said:
This is worthy of a God-Like post...



This was posted by JumpingJack at the link Turpit provided.


So what he's saying is you wouldn't even consider a crash to be form the L3 cache, you'd probably suspect Windows and just simply hit the reset button on your PC. TLB bug is over-rated at best.

From the users posting there it would appear as though Phenom is just fine. Imagine that.


I agree.

Every indication I get is that the Bug is something rarely seen if never. In short, I think the biggest mistake AMD made was to mention that it existed. lol
January 6, 2008 6:50:12 PM

It really depends on what you do with your current PC. If it's a lot of encoding, rendering of image editing work, an upgrade to Phenom would be useful. However, if you just mainly game, stick to your 5000+ BE.
January 6, 2008 7:23:25 PM

I didn't tell him to wait for B3 because of the TLB bug, but because B3 should clock higher than the current revision - which, let's face it, is a pretty sad overclocker.
January 6, 2008 7:49:58 PM

If you don't OC maybe a phenom is ok, but if you do, the Q6600 will hit it hard.

That thread at xtremesystem is oriented to those who haven't flashed their bios to a TLB-bug-correction one. I wouldn't mess without the correction in my case but who knows.

There's also this TLB-bios vs No-TLB-bios benchmark at http://techreport.com/articles.x/13741

Unluckly there's no benchmark out there that shows heavy multitasking as the famous toms hardware "multitasking test" a year or 2 ago, where the firsts Athlons 64 X2 had problems to catch the dual core Intels that had HyperThreading (when 4xApp at full throttle were running).

What I would like to see is how does the Phenom improves when you put 4 memory modules, where each core can access each module independently. Also, I would like to know if an old AM2 mobo can support DDR2 1066 when sporting a phenom. I think it should be possible but not sure...

I'm actually on an asrock 754 K8upgradeNF3 + the AM2 module (have the 939 module also). As the mobo¡ sports a small 2Mbit bios, It doesn't support a B2 phenom, but I'm pretty sure that without the TLB bug, the bios will be small enough to fit on a 2Mbit bios and then all motherboards may be upgradable, I suppose...
January 6, 2008 8:06:47 PM

My asus crosshair will do 1066 with my ballistix tracer dimms.
They are even better than my more expensive corsair pc8500 dimms ironically.

Independent dimm access for each core on a quad core would be a great idea providing the multithreaded approach can be taken with everything. I think there would be a nice gain to be had.

Additionally, the thing with the backwards AM2 compatibility that everyone blames on AMD is not really AMD's fault. It's the failure of board manufacturers to update the BIOS for their boards. I believe all AMD really implied is that the AM2 architecture would support it.
We have had plenty of BIOS issues regarding chip support on many platforms in the past.

Asus boards seem to have no issues with this though which none of the naysayers seem to give credit for.

As soon as the new stepping hits the market I am there, and hopefully by then ASUS will have a new ROG board for it too with the new AMD chipset. I love Asus ROG boards. I also have hope for the "spider" platform in the future I wish these manufacturers would include crossfire AND SLI support in these boards too. That kind of urks me that you don't have a choice what cards to use when you want upscale.

I would really like to take advantgage of the excellent Radeon offerings right now, but can't to my knowledge.
January 6, 2008 8:29:26 PM

righteous said:
I agree.

Every indication I get is that the Bug is something rarely seen if never. In short, I think the biggest mistake AMD made was to mention that it existed. lol


Do you remember the stink a few years ago about the Ford Explorer rolover tests? When the testers did a standrd Jturn the Explorer tipped over 5 out of 12 times, while the competiton kept the rubber side down and the shiny side up. Hey, I don't go down the highway at 50 mph and suddenly yank hard on the wheel but damn it, I'm not gonna buy the vehicle that tips over when you do! And I'm also not buying a chip with an acknoledged major flaw. Just like I'm also not buying the comments that it doesn't apear in everyday use. It's there. It has to be serious enough to warant a new steping. I'm a consumer. I have a choice. I'm buying anything but that.
January 6, 2008 8:50:27 PM

the TLB fix is not about CPU load (encoding comment) its mostly related to the VMware type programs that use the CPUs virtualization (that needs to be set to on in BIOS) so turn off virtualization in bios and it be fine, i been testing this 9500 phenom with FAH smp and single clients and not crashed once (thats useing an asrock mobo that has an 20 pin ATX power cable not an 24 one)
any one who has got an AMD chip set i recommend you install overdrive and move the slider bar to faster (so it makes the green dot at the top right go yellow or red) that will turn off the TLB fix

for none AMD chip set and do not have an option to turn off micro code updates, you got to live with the fix or use an bios that does not have it
January 6, 2008 9:18:25 PM

OlSkoolChopper said:
Do you remember the stink a few years ago about the Ford Explorer rolover tests? When the testers did a standrd Jturn the Explorer tipped over 5 out of 12 times, while the competiton kept the rubber side down and the shiny side up. Hey, I don't go down the highway at 50 mph and suddenly yank hard on the wheel but damn it, I'm not gonna buy the vehicle that tips over when you do! And I'm also not buying a chip with an acknoledged major flaw. Just like I'm also not buying the comments that it doesn't apear in everyday use. It's there. It has to be serious enough to warant a new steping. I'm a consumer. I have a choice. I'm buying anything but that.


The difference between the Ford Explorer and the AMD Phenom is this: The Explorer had recurring results. The Phenom has had, well, none that I have actually heard of. Only place I've EVER heard the TLB bug occur at was in AMD's test labs. Other than that, it hasn't happened to anyone as far as I know.

January 6, 2008 9:31:36 PM

justinmcg67 said:
The difference between the Ford Explorer and the AMD Phenom is this: The Explorer had recurring results. The Phenom has had, well, none that I have actually heard of. Only place I've EVER heard the TLB bug occur at was in AMD's test labs. Other than that, it hasn't happened to anyone as far as I know.



Somewhere in the forums there is a link buried to a french sites article. The article concerns the Tahoe testing. During that test, while the AMD 'overseer's back was turned, the authors overclocked their rig, and supposedly (if I recall correctly) encountered the bug. They may or may not have encounterd the bug, or a different problem.

There are a few individuals on various sites who are looking to duplicate the bug. The results they have encountered are inconclusive. Those who have encounterd problems with their systems are unsure if the crashes were caused by the TLB bug, hardware, OS or drivers. Google "TLB bug occurance" or TLH bug crash" to find the links to the forums were the testers discuss there results.

As was pointed out, this bug is supposedly (per AMDs press release) specific to virtualization, but as AMD has not released the exact data (hardware, config, setup, software, environement) under which the bug occurs, those trying to duplicate the bug have had to experiment, and thus are unsure of their results. However, as has been stated, the probability of occurance with the 9500/9600 Phenoms appears small.
January 6, 2008 10:19:07 PM

turpit said:
Somewhere in the forums there is a link buried to a french sites article. The article concerns the Tahoe testing. During that test, while the AMD 'overseer's back was turned, the authors overclocked their rig, and supposedly (if I recall correctly) encountered the bug. They may or may not have encounterd the bug, or a different problem.

There are a few individuals on various sites who are looking to duplicate the bug. The results they have encountered are inconclusive. Those who have encounterd problems with their systems are unsure if the crashes were caused by the TLB bug, hardware, OS or drivers. Google "TLB bug occurance" or TLH bug crash" to find the links to the forums were the testers discuss there results.

As was pointed out, this bug is supposedly (per AMDs press release) specific to virtualization, but as AMD has not released the exact data (hardware, config, setup, software, environement) under which the bug occurs, those trying to duplicate the bug have had to experiment, and thus are unsure of their results. However, as has been stated, the probability of occurance with the 9500/9600 Phenoms appears small.



Exactly. There is no conclusive evidence to show that the bug was or was not a direct result of overclocking. I've had many system crashes, all of which occurred when I overclocked my system. I personally think it was due to overclocking and NOT the TLB bug. If you overclock a system, the first thing you due is compromise system stability. If anything they just proved that OCing does compromise stability, and I do not think they showed that it was Phenom and the L3. This is my own view of course, take it for what you want.

However, granted XtremeSystems posts, from actual people using the chip, they have no been able to replicate the bug. It should also be noted that people TRYING to replicate the bug, can in fact, not replicate it. So you obviously have to try very hard to get it to occur. Thus the bug occurance is extremely slim and has been over-hyped beyond belief.
January 6, 2008 10:47:47 PM

if the bug was related to SSE or so, then it would have been significative. But as long as it only affects virtualization then everyone has to turn off the fix... because I don't think anybody is using that.

My question is, it there an option to disable L3 altogether? I know it sound somewhat stupid but it would be an interesting bench to see the impact of L3.
January 7, 2008 1:16:29 AM

justinmcg67 said:
Exactly. There is no conclusive evidence to show that the bug was or was not a direct result of overclocking. I've had many system crashes, all of which occurred when I overclocked my system. I personally think it was due to overclocking and NOT the TLB bug. If you overclock a system, the first thing you due is compromise system stability. If anything they just proved that OCing does compromise stability, and I do not think they showed that it was Phenom and the L3. This is my own view of course, take it for what you want.

However, granted XtremeSystems posts, from actual people using the chip, they have no been able to replicate the bug. It should also be noted that people TRYING to replicate the bug, can in fact, not replicate it. So you obviously have to try very hard to get it to occur. Thus the bug occurance is extremely slim and has been over-hyped beyond belief.


According to AMD, the higher clocked 2.4GHz Phenom 9700's were pulled from the launch due to the translation lookaside buffer errata.
http://www.pronetworks.org/index.php/technology-news/197
Quote:
We asked AMD to provide us with official statement on this huge problem, and we got a reply back from the firm: “The Translation Lookaside Buffer (TLB) errata is an L3 protocol issue causing a system hang when running certain client workload applications independent of platform. AMD is immediately introducing an updated BIOS which will correct the TLB errata”.

This problem was found during speed-binning the B2 revision processors, and this was the cause for the Phenom FX 3.0 GHz delay. It turns out that some CPUs running at 2.4 GHz or above in some benchmarking combinations, while all four cores are running at 100% load, can cause a system freeze.


According to the information provided at the time of release, the problem is reported to occur only in processors at 2.4GHz or above. As such, it should not affect the 2.2/2.3GHz parts currently available through retail. Assuming this information is accurate and not indicative of clockspeed binning or heat saturation problems.

The individuals I have read that are looking to replicate the bug have done so on overclocked systems in the 2.6-2.7GHz range. One individual was able to successfully push his Phenom as high as 2.8HHz(2.77) and during his tests under varying configurations, he was able to account for any difficulties he had with one exception, with 3 cores loaded on VM with the update. In that one instance he could not identify the problem, and therefore could not discount the possibility of the TLB bug. Running the same configuration without the patch, the problem did not occur.

So the current 9500/9600s should be safe from the TLB bug and, as a bonus able to run without the performance penalizing fix, however, OCing may lead to problems.

As for the over hype, I would say this stems from distrust of AMD rather than any malevolence towards them...a few trolling Intel phanboys aside. It would seem there are those (myself included) who simply do not take AMD at their word any longer, especially after the Tahoe fiasco. Now that private individuals actually have retail Phenom systems, are testing them and are unable to conclusively duplicate the TLB bug, we have some real world proof that the Phenom 9500 and 9600s are in fact unlikely to experience this problem. Of note however, is that most of the individuals attempting to duplicate the TLB bug have had their systems a relatively short time, so while the results are promising, I would not yet call them conclusive. The results look good, but a few more weeks and a few more 'real' users should provide reasonably conclusive proof.



January 7, 2008 2:43:45 AM

justinmcg67 said:
The difference between the Ford Explorer and the AMD Phenom is this: The Explorer had recurring results. The Phenom has had, well, none that I have actually heard of. Only place I've EVER heard the TLB bug occur at was in AMD's test labs. Other than that, it hasn't happened to anyone as far as I know.


Plus, though he said he'd never buy a CPU with a known bug, I"m quite sure he's bought Intel CPU's with known bugs. They have errata too.

Fry's just had a special Friday January 4th bundling the 2.3 gigahertz Phenom and an ASUS M2A-VM board for $274. That's the board that I have. The motherboard at Fry's should already have the correct bios.

So, I'll have to flash the bios on our two boards before I get a Phenom. I'm just not sure if I should wait for 45nm before I upgrade. When is that arriving? Right now, I could probably survive with an Athlon X2 3800+ and an Athlon X2 4600+.

I'm also wondering what the pricing on the triple cores will be, and what wattage the 45nm Phenoms will have. I'd hate to buy now and then see a triple core for $129 with both triple and quad cores at 65watts.


January 7, 2008 2:55:02 AM

turpit said:
Somewhere in the forums there is a link buried to a french sites article. The article concerns the Tahoe testing. During that test, while the AMD 'overseer's back was turned, the authors overclocked their rig, and supposedly (if I recall correctly) encountered the bug. They may or may not have encounterd the bug, or a different problem.


I remember you asking me about the French site that messed up the Tahoe test and neither one of us could find it at the time. Ok, I could put it simply that if something could be messed up, the French could find a way to do it.

That aside, I asked my insurance agent about the Phenom, as I do a some medical research. His reply was simple, I get a Phenom and I loose my liabilty insurance; as in, no malpractice insurance, no nothing. It wouldn't matter if there was ever a crash or not, if it ever gave out some bad information or not. All that a courtroom would care about is that it "could have happened", and I'd be screwed. So, will I buy a Phenom before the B3 stepping comes out and its certified not to have the TLB errata? Sure I will; when hell freezes over, pigs fly, and the cow really jumps over the moon.

As for Jack's article in the other web site, I've known Jack for a few years, respect him and his work and I agree with his conclussions about the Phenom concerning the errata. But I can't afford the chance of the present Phenom. My biggest gripe with Phenom isn't as much the processor as the lies that were told about it. I am left with the feeling that no matter what AMD may say about it, I can't trust AMD's word on the subject, or about most anything else at this point.

I hope the B3 will get everything fixed and running properly. I hope the motherboard problem with the SB600 south bridge will be resolved with the change to a SB700 south bridge. But until those problems are solved, there will be no Phenom in my house.
January 7, 2008 1:06:18 PM

yipsl said:
Plus, though he said he'd never buy a CPU with a known bug, I"m quite sure he's bought Intel CPU's with known bugs. They have errata too.


Damn near every CPU has errata. Its a mater of magnitude. I woudlnt have bought any of the old Intels with the math problem even after the "fix" either. If an errata is so huge as to require a preanounced revision many months in the future, then thats big enough to scare me away. You can flip any car on its side, even Ferrari's new F1, if you try hard enough. The question you have to ask is: "does this car have a greater than average probibility of fliping over in more or less normal driving." The Explorer did. So does Phenom. Both of them can go to hell as far as my checkbook is concerned. Aint gettin any of my hard earned cash.

sailer said:
That aside, I asked my insurance agent about the Phenom, as I do a some medical research. His reply was simple, I get a Phenom and I loose my liabilty insurance; as in, no malpractice insurance, no nothing. It wouldn't matter if there was ever a crash or not, if it ever gave out some bad information or not. All that a courtroom would care about is that it "could have happened", and I'd be screwed.
\

Yikesdiyikesdeeyikes!!! If the insurance industry is already up on the Phenom bug, and they're telling people that it will void their policies, this is way more serious than even I thought!
January 7, 2008 2:54:28 PM

OlSkoolChopper said:

Yikesdiyikesdeeyikes!!! If the insurance industry is already up on the Phenom bug, and they're telling people that it will void their policies, this is way more serious than even I thought!


I can't say every insurance agency is concerned with this, only mine and because I do some medical research. Actually, I can't even say if the agent was over reacting at the time. You have to keep in mind that in the good old USA, people can sue for almost anything and find an ignorant jury to believe them. I have a doubt that the errata bug is all that bad. My biggest complaint is how slow the chip is and the problems in overclocking it.
January 7, 2008 3:30:11 PM

sailer said:
I can't say every insurance agency is concerned with this, only mine and because I do some medical research. Actually, I can't even say if the agent was over reacting at the time. You have to keep in mind that in the good old USA, people can sue for almost anything and find an ignorant jury to believe them. I have a doubt that the errata bug is all that bad. My biggest complaint is how slow the chip is and the problems in overclocking it.


Based on insurance and the fact that you do medical research you claim you are worried about the TLB errata that at worst could freeze your system.

However, you ARE willing to overclock your system?

Danger Danger Will Robinson! Danger. Something is awry with this logic.

EDIT: Just in case the Will Robinson reference is too old for you perhaps I'd better spell it out specifically. Based on what you posted I must doubt your veracity.
January 7, 2008 4:13:32 PM

keithlm said:
EDIT: Just in case the Will Robinson reference is too old for you perhaps I'd better spell it out specifically. Based on what you posted I must doubt your veracity.


I have three computers, one for gaming and two for my office. The office computers never, ever get overclocked. Only my gaming computer gets overclocked. As my gaming computer ages, it gets turned into an office computer after all the games and such are uninstalled. I then build a new gaming computer. My insurance agent knows this, that a Phenom powered gaming computer this year would turn into an office computer in a year or two. Thus his warning about the Phenom.

At 56 years old, I remember Will Robinson and Lost in Space.
January 7, 2008 4:30:06 PM

I remember the impossibly crappy movie. I wish i didnt tho!
January 7, 2008 4:32:54 PM

OlSkoolChopper said:
I remember the impossibly crappy movie. I wish i didnt tho!


Yes, the movie fit into the Oscar nomination catagory of "Movies that should never have been released".
January 7, 2008 4:36:31 PM

Gahhh... Dr. Smith. Hmmmm, would he kinda be like.. Hector? :oops: 
January 7, 2008 4:52:59 PM

Let me think. Dr. Smith turned into a Spider, Hector is trying to sell us Spiders. :??:  Is this co-incidense? :o 
January 7, 2008 4:56:05 PM

I was thinking of the older series of lost in space... Didn't really think about the movie production film till ya mentioned it. That is... scary. :cry: 
!