Phenom VS. Intel Quadcores (Q6600) ?

So what is the point that you wanted to make with this long errata argument?

To bring out the facts.

Tell me how many of those Dell, HP, and other OEM system users that have actually updated their BIOS?

Does this means that someone, who bought an OEM C2D system in the early days before the microcode update was out and he has never flashed the BIOS before, is running a computer with a bigger risk than someone who is running a Phenom without the fix?

No, none of the C2D errata are as severe as the Phenom one. They are mostly 'cosmetic' patches... where stability issues are mostly theoretical, or at least related only to certain OS/driver code.
So yes, an unpatched C2D system is less stable than a patched C2D system... But that doesn't make it as unstable as an unpatched Phenom system.
Why not? Well, the OEMs actually refused to sell Opteron systems because they encountered stability issues. The Xeon systems have undergone the same heavy duty test cycle, but they didn't encounter any problems, and the servers were launched on schedule.
So whatever errata the C2D may have, they were never encountered 'in the wild'.
 
Funny how your "yes" turns to "no" Scali :p

Yes, I read it wrong, you said less severe, not more severe.
The rest of the post doesn't make sense with 'yes', it does with 'no'.

Shame you had to waste posts on that.
 
To bring out the facts.



No, none of the C2D errata are as severe as the Phenom one. They are mostly 'cosmetic' patches... where stability issues are mostly theoretical, or at least related only to certain OS/driver code.
So yes, an unpatched C2D system is less stable than a patched C2D system... But that doesn't make it as unstable as an unpatched Phenom system.
Why not? Well, the OEMs actually refused to sell Opteron systems because they encountered stability issues. The Xeon systems have undergone the same heavy duty test cycle, but they didn't encounter any problems, and the servers were launched on schedule.
So whatever errata the C2D may have, they were never encountered 'in the wild'.
What are the facts? Are the bold part of your post facts? I still don't get what is the point that you want to make with your argument?
 
What are the facts? Are the bold part of your post facts? I still don't get what is the point that you want to make with your argument?

Yes, they are facts.
The point is the truth. Death to fanboys, death to lies, death to misinformation, death to FUD. Truth FTW!
 
Yes, they are facts.
The point is the truth. Death to fanboys, death to lies, death to misinformation, death to FUD. Truth FTW!

Can you back up your facts?
What is the truth? Where is the misinformation? Is saying "the bug is so rare that even with the fix disabled, it won't affect most user and shouldn't be much of a concern for most user" is not the truth? Is this misinformation?
 
This article says that it's during the validation done by OEMs that the bug was discovered, and that the bigger OEMs refuse to sell the buggy chips because of the performance loss with the fix: http://core.tweakers.net/reviews/789/2/jaaroverzicht-amd-en-de-barcelona-soap-pagina-2.html

Is saying "the bug is so rare that even with the fix disabled, it won't affect most user and shouldn't be much of a concern for most user" is not the truth? Is this misinformation?

In a way it's misinformation, or rather misleading information, because it's a very vague statement. It's nothing more than an opinion, not based on any quantitative analysis or fact.
Words like 'rare', 'most users', 'much of a concern' really don't say anything, and leave it all up to the reader's interpretation.
Where are the facts that lead to this opinion of yours?

As I already said... 6 out of 10 is still 'most users', but that would leave 4 really unhappy users out of 10, which is a lot.
I've already addressed the 'rare' and 'concern' issues aswell earlier.
 
Firstly don't you want to back up your facts above (OEM refusal and unstable Phenom)?

I did, had to find the article where I read it. It's in Dutch though, but that's not my problem.

Here they are:

They are facts, but they don't give substance to the hollow terms 'rare', 'most users' or 'much of a concern'.
You really don't know HOW rare the bug is, do you? You've not seen it... Is that just your dumb luck, or perhaps it's not happened yet because it's more rare than the few weeks you could possibly have used a K10 so far... but not rare enough to not happen ever during the ownership of a K10.
As for 'most users'... you quote a handful of Phenom owners on a forum... I doubt they'd quantify for 'most users'. In fact, some people actually report that they have managed to reproduce the bug. Factor those in aswell.
As for 'much of a concern'... If you don't encounter the bug, you don't know what the consequences will be... As I say, it's entirely possible that your entire HDD is wiped out. I think that WILL be much of a concern if that happens to an unlucky user. Even a crash at the wrong time could be a big concern.

So my conclusion on the above facts would be way different: Don't buy the CPU, you don't know what will happen without the fix, and the performance with the fix installed is just too disappointing to consider buying the chip at all.
 
I did, had to find the article where I read it. It's in Dutch though, but that's not my problem.



They are facts, but they don't give substance to the hollow terms 'rare', 'most users' or 'much of a concern'.
You really don't know HOW rare the bug is, do you? You've not seen it... Is that just your dumb luck, or perhaps it's not happened yet because it's more rare than the few weeks you could possibly have used a K10 so far... but not rare enough to not happen ever during the ownership of a K10.
As for 'most users'... you quote a handful of Phenom owners on a forum... I doubt they'd quantify for 'most users'. In fact, some people actually report that they have managed to reproduce the bug. Factor those in aswell.
As for 'much of a concern'... If you don't encounter the bug, you don't know what the consequences will be... As I say, it's entirely possible that your entire HDD is wiped out. I think that WILL be much of a concern if that happens to an unlucky user. Even a crash at the wrong time could be a big concern.

So my conclusion on the above facts would be way different: Don't buy the CPU, you don't know what will happen without the fix, and the performance with the fix installed is just too disappointing to consider buying the chip at all.
Wow, that is all that I can say.

Can you tell for a certain when will a component fail on you? Forget about your entire HDD being wiped out by the bug, I would be more concerned with the HDD itself failing. Forget about a crash at the wrong time, a power outage during critical timing would more of a concern for me. Even when using an UPS, I would still be concerned with a failing PSU that can take other hardwares as well.
 
Can you tell for a certain when will a component fail on you? Forget about your entire HDD being wiped out by the bug, I would be more concerned with the HDD itself failing. Forget about a crash at the wrong time, a power outage during critical timing would more of a concern for me. Even when using an UPS, I would still be concerned with a failing PSU that can take other hardwares as well.

All that can be solved with redundancy. Any half-decent mission-critical system will have a redundant PSU and a RAID mirror HDD setup. So virtually any kind of component failure can be taken care of... add to that hot-pluggability, and you have yourself a pretty much failsafe computing environment.
So well, for server systems, the only choice is to have the TLB fix running, but because of the performance impact (especially with VM-related tasks), this makes the CPU bad value for money, which is why the big OEMs refuse to sell it.

Now most desktop users won't require this level of reliability. However, PSUs, HDDs and various other hardware have well-defined reliability specifications. The manufacturer will just quote you with an MTBF and some other info, so you know more or less what to expect, and you can replace the hardware long before you will enter the 'danger zone'. Namely, it's a wear-and-tear issue. The hardware will gradually degrade over time.
With the addition of SMART on HDDs, you will usually be warned long before the HDD will actually start to lose data.
So yes, there is a risk that the hardware may fail, but it is a 'controlled risk'. And you can do scheduled backups etc to keep your data safe.

Back to the TLB issue... Firstly we have NO data on how often the bug occurs... So far we know that some people have been using the CPU for a few weeks... that's all we can go on.
Secondly, there's no way to predict when the bug is going to strike. You can't just say "Oh, today is the day the bug will strike... I'll just enable the fix, and tomorrow I'll be good, and disable it again".
Thirdly, there's no way of telling how the bug will manifest itself. You could have silent data corruption... So even when you are backing up your data, there's no way to guarantee that your documents were actually saved properly the last time you used them, or that they will be backed up properly.

It's just a much bigger risk than using a system with only one PSU and no mirrored RAID. It's not as predictable.
I would never take the risk, and I wouldn't recommend it to anyone else either, if they are at all concerned about their data. I can imagine that if you use your PC only as a gaming console, you might want to run without the fix. You can always just reinstall your games and such, worst that could happen is that you'll lose your savegames.
But if like me, you use valuable documents and sourcecode on your system, you can't really take the risk of any of that corrupting, can you?
Now I can take a random crash, a failing harddrive or PSU or whatever, because I keep backups and all... But the risk that my documents or backups are useless because my CPU decided to bug without telling me about it... no, I'm not going to take it. It would cost WAY more than just investing in a slightly more expensive, but perfectly reliable Q6600 system.
 
So what are you trying to say Scali2? Do you want to meet every single Phenom owner and ask them to test certain things on their computer to see if the bug manifests? If 'rare' and 'much of a concern' is bothering you, maybe you should take the initiative to exactly see how often this bug appears. As far as Im concerned, you are just beating around the bush and making such a big deal out of something that seems insignificant for the regular consumer, Period.

I'm just trying to inform people about the risks, because this thread makes it painfully obvious that most of you haven't the slightest idea what kind of bug this is, and what the consequences will be when it strikes.
I think everyone needs to have a clear idea about the consequences before they decide to buy a 'risky' CPU such as this one.
And for the ones who already bought one, they should be aware of the consequences of running it without fix aswell.

Is the notion of having a buggy CPU bothering you so much? Face the facts, there is no 100% bug-free CPU. Else, I suggest you wait an eternity for one to come out.

I know that. What you and all those other **** in this thread don't realize is the difference in *severity* of the bugs.
There have been very few CPUs in the history of computing that had a bug as severe as this one. The closest example that comes to mind is the Pentium FPU bug. Ofcourse Intel did a total recall on that one, much better than this performance-sucking TLB fix.


What do you want to say with that post?
Haven't you people ever been to school? Don't you know anything about scientific research and statistics? Again I would like to point you to Popper.

Just stop arguing when you **** don't have any idea what you are talking about. Go ask someone else who really understands the bug, such as dexvx, and they'll say the same as me.
All you guys are doing is arguing that I'm wrong when the facts and explanations are right there, staring in your face. You just don't understand them. That's just really dumb, if you ask me.
You're basically insulting and harassing me because you don't want to believe what I'm saying, and are not capable of understanding it yourself and making your own analysis of the situation. You're just borrowing third-party opinions, and you happen to select the one that suits your agenda. I have no other word for it but dumb. People like you don't belong on a tech forum. You're not talking about technology, you are just about emotional hangups about brands and products, not backed up by facts, but by irrational 'gut feelings'. Go away.
 
All that can be solved with redundancy. Any half-decent mission-critical system will have a redundant PSU and a RAID mirror HDD setup. So virtually any kind of component failure can be taken care of... add to that hot-pluggability, and you have yourself a pretty much failsafe computing environment.
So well, for server systems, the only choice is to have the TLB fix running, but because of the performance impact (especially with VM-related tasks), this makes the CPU bad value for money, which is why the big OEMs refuse to sell it.

Now most desktop users won't require this level of reliability. However, PSUs, HDDs and various other hardware have well-defined reliability specifications. The manufacturer will just quote you with an MTBF and some other info, so you know more or less what to expect, and you can replace the hardware long before you will enter the 'danger zone'. Namely, it's a wear-and-tear issue. The hardware will gradually degrade over time.
With the addition of SMART on HDDs, you will usually be warned long before the HDD will actually start to lose data.
So yes, there is a risk that the hardware may fail, but it is a 'controlled risk'. And you can do scheduled backups etc to keep your data safe.

Back to the TLB issue... Firstly we have NO data on how often the bug occurs... So far we know that some people have been using the CPU for a few weeks... that's all we can go on.
Secondly, there's no way to predict when the bug is going to strike. You can't just say "Oh, today is the day the bug will strike... I'll just enable the fix, and tomorrow I'll be good, and disable it again".
Thirdly, there's no way of telling how the bug will manifest itself. You could have silent data corruption... So even when you are backing up your data, there's no way to guarantee that your documents were actually saved properly the last time you used them, or that they will be backed up properly.

It's just a much bigger risk than using a system with only one PSU and no mirrored RAID. It's not as predictable.
I would never take the risk, and I wouldn't recommend it to anyone else either, if they are at all concerned about their data. I can imagine that if you use your PC only as a gaming console, you might want to run without the fix. You can always just reinstall your games and such, worst that could happen is that you'll lose your savegames.
But if like me, you use valuable documents and sourcecode on your system, you can't really take the risk of any of that corrupting, can you?
Now I can take a random crash, a failing harddrive or PSU or whatever, because I keep backups and all... But the risk that my documents or backups are useless because my CPU decided to bug without telling me about it... no, I'm not going to take it. It would cost WAY more than just investing in a slightly more expensive, but perfectly reliable Q6600 system.
So you never had any hardware failure without a warning? Yet you said that you are a long time computer user. Are sure that your Sata controller is that reliable when doing your backup? Do you get any warning before your RAM fails? Btw what most desktop user? 6 out of 10? What level of reliability?

You talked about facts but in reality your never provided facts to backup your claims(I'm still waiting your OEM refusal facts). Saying anything could happen is far from scientific.

W4ngsta is right, you are just beating the bush. Making thing looks bigger than it really is.
 
So you never had any hardware failure without a warning?

Not in the past 5 years or so, no.
Could be luck, could be the fact that I don't buy stuff because it's the cheapest, but rather select stuff that's reliable and 'comfortably specced', and am willing to pay a bit extra for it.
I'm just not taking risks when it comes to buying motherboards, PSUs and such. I buy stuff that's built to decent quality, has good diagnostics and protection, and has plenty of headroom for what I will be doing with it.

I even have some 386, 486 and Pentium systems that still run, and my main home server is a PII-333 which runs overclocked to 350 MHz 24/7. The system is over 10 years old now. It uses a relatively new HDD ofcourse, I wouldn't trust a 10 year old HDD. The PSU is also from a newer system.
It's very reliable actually:
scali@scali~$ w
1:25PM up 221 days, 19:53, 1 user, load averages: 0.00, 0.00, 0.00

Basically hasn't crashed since I installed and configured it. You can reach it at http://scali.eu.org

So hardware failures, on decent quality parts? Very rare indeed, haven't had them in YEARS, and I'm running some old shit here :p

Yet you said that you are a long time computer user. Are sure that your Sata controller is that reliable when doing your backup? Do you get any warning before your RAM fails?

Don't need to. The controller is covered by CRC checks and such, so even if it isn't reliable, you'll know about it. You know where and when it happens, and what data is and isn't corrupted.
As for RAM failing... Never happened to me. The MTBF on ram chips is 'stupid high'. Something like once in a million years or so. It's all specified by the manufacturer anyway, as I already stated. But if I was worried about that, I could opt for ECC memory.

Btw what most desktop user? 6 out of 10?

Depends on your definition of 'desktop user'.
Technically I would say 'all' desktop users. Because I'd say they would have to be workstation users if they DID require that kind of reliability... If they're 'just' desktop users, they would probably get that reliability because they are working with network storage or application servers, which run on servers that solve the reliability problems for them. Their own desktop would just be an empty shell, and could crash without any impact on the data integrity.
I'm just leaving some room because I know not everyone uses their PC in this way, even though from a strategic point-of-view, they should. The gray area between desktop and workstation class computers, so to say.

What level of reliability?

The level of reliability I described in the paragraph above (redundancy, RAID, hot-pluggability, etc).
Should have been painfully obvious. It is clear to me that you don't even bother to understand what I'm writing, and just take things out of context to harass me.
This remark is just ridiculous. Practice some reading comprehension.

Oh, and I *did* provide facts... Read up, the tweakers.net link. I expect yet another apology for falsely accusing me.
 
So what are you trying to say Scali2? Do you want to meet every single Phenom owner and ask them to test certain things on their computer to see if the bug manifests? If 'rare' and 'much of a concern' is bothering you, maybe you should take the initiative to exactly see how often this bug appears. As far as Im concerned, you are just beating around the bush and making such a big deal out of something that seems insignificant for the regular consumer, Period.

Is the notion of having a buggy CPU bothering you so much? Face the facts, there is no 100% bug-free CPU. Else, I suggest you wait an eternity for one to come out.

And for the record, read this: http://www.xtremesystems.org/forums/showpost.php?p=2654041&postcount=27

But that's exactly it. This discussion is focused on the severity of the bug, which for the OEMs, is too much for them to bare. You must understand that even though rare, this kind of bug cannot be present, otherwise, if it occurs, the OEMs will be held responsible and will need to replace the hardware or provide a full compensation. That's bad rep right there and a risk that no money making company, can take. So yes, regular Joe may not be impacted and not even the big customers, that these OEMs supply, but the risk is there and they just can't take it. Can you imagine having to replace or compensate for 1000 or more workstations, because the bug was reported in just one of them ? Extremely bad rep and probably reason enough to end the contract.
 
But that's exactly it. This discussion is focused on the severity of the bug, which for the OEMs, is too much for them to bare. You must understand that even though rare, this kind of bug cannot be present, otherwise, if it occurs, the OEMs will be held responsible and will need to replace the hardware or provide a full compensation. That's bad rep right there and a risk that no money making company, can take. So yes, regular Joe may not be impacted and not even the big customers, that these OEMs supply, but the risk is there and they just can't take it. Can you imagine having to replace or compensate for 1000 or more workstations, because the bug was reported in just one of them ? Extremely bad rep and probably reason enough to end the contract.

What are you on? There is a fix available for OEMs too.

Btw to Scali2, you said that OEM refused to sell opteron systems, where is your link to back it up?
 
ALL CPU have errata that can trigger in very obscure situations. The vast majority of erratum in CPUs are left unfixed, since they are deemed VERY rare or simply aren't significant enough to warrant a fix. However, it does appear the TLB errata happens to be more severe than your average errata.

Thats mostly false. Most errata left in production CPU's are cosmetic, meaning they have no impact on operations. Then there are errata that is caused by a series of events that cannot happen - which is quite different than rarely happening. Very few can actually be reproduced with commercial software at any level (which is very bad).

What are you on? There is a fix available for OEMs too.

Btw to Scali2, you said that OEM refused to sell opteron systems, where is your link to back it up?

You can no longer configure K10 Opterons via Dell or HP as of now. It seems to me that the problem for servers is serious enough that Dell and HP (and I'd imagine other vendors) basically issued a stop order for K10 Opterons. They are most likely waiting for a new stepping that is coming out Q2 2008.
 
What are you on? There is a fix available for OEMs too.

Btw to Scali2, you said that OEM refused to sell opteron systems, where is your link to back it up?

You missed my point. I was talking about OEMs even thinking of looking beyond the bug, just because it's so rare. They can't afford that.
 
You can no longer configure K10 Opterons via Dell or HP as of now. It seems to me that the problem for servers is serious enough that Dell and HP (and I'd imagine other vendors) basically issued a stop order for K10 Opterons. They are most likely waiting for a new stepping that is coming out Q2 2008.

Scali2 claims that OEMs refused to sell the chips. I asked him to back it up but he came up with nothing yet because from what I've read before, it seems like AMD itself that didn't ship those chips. Here are the links:
http://www.crn.com/hardware/204701424
http://www.techreport.com/discussions.x/13721
http://www.theregister.co.uk/2007/12/06/opteron_delays_amd/
http://news.softpedia.com/news/AMD-039-s-Barcelonas-no-Pit-Stop-but-Ship-Stop-73003.shtml

AMD man Phil Hughes told InternetNews, "It's only a stop ship if it's shipping in volume, and we're only shipping Barcelona for specific customer commitments, like larger volume deployments."

Quite logical actually, when you know that your current product has a problem and you have a limited resource, it doesn't make any sense to take orders and carry on with the production, it is better to stop the production and concentrate your limited resource into your next product so that you can release it in high volume.
 
You can no longer configure K10 Opterons via Dell or HP as of now. It seems to me that the problem for servers is serious enough that Dell and HP (and I'd imagine other vendors) basically issued a stop order for K10 Opterons. They are most likely waiting for a new stepping that is coming out Q2 2008.

Exactly. You think it's AMD's choice? Ofcourse not!
The OEMs can only sell a server system with the fix enabled. And that's where the problem is.
The OEMs originally did their validation with the non-fixed platform. The CPUs failed this validation... then it was back to the drawing board for AMD... AMD came up with the TLB fix. However, the OEMs are not taking chances... So they will AGAIN have to go through the validation process before they can sell systems with the TLB fix.
Now the problem is that the TLB fix degrades performance, and it was available far too late.
So the OEMs have decided not to even bother with this fix at all. They will wait for the later stepping, and then they'll validate that one and start selling it, if it validates okay.
So it is these OEMs that caused AMD to stop shipping Opterons.
Then AMD ofcourse wants to tell the media that *they* found the bug, and that *they* decided to cancel shipment... But this is just their spindoctors at work. Dell, HP and other major OEMs tell a different story.
Besides, AMD isn't all that credible when they stop shipping Opterons, but still sell Phenoms with the exact same bug. Do any of the major OEMs sell any Phenoms by the way? I don't think I've seen any Dells or HPs with Phenom yet either.
OEMs just face the big problem that they can't sell systems with the fix disabled because of potential stability problems... and they can't sell them with the fix enabled because the performance wouldn't be good value for money. The Spider platform is not very attractive to OEMs at this point.

I don't know why tweakers.net is one of the few sites that actually dares to mention the truth... Perhaps other sites 'depend' on AMD too much.
 
Exactly. You think it's AMD's choice? Ofcourse not!
The OEMs can only sell a server system with the fix enabled. And that's where the problem is.
The OEMs originally did their validation with the non-fixed platform. The CPUs failed this validation... then it was back to the drawing board for AMD... AMD came up with the TLB fix. However, the OEMs are not taking chances... So they will AGAIN have to go through the validation process before they can sell systems with the TLB fix.
Now the problem is that the TLB fix degrades performance, and it was available far too late.
So the OEMs have decided not to even bother with this fix at all. They will wait for the later stepping, and then they'll validate that one and start selling it, if it validates okay.
So it is these OEMs that caused AMD to stop shipping Opterons.
Then AMD ofcourse wants to tell the media that *they* found the bug, and that *they* decided to cancel shipment... But this is just their spindoctors at work. Dell, HP and other major OEMs tell a different story.Besides, AMD isn't all that credible when they stop shipping Opterons, but still sell Phenoms with the exact same bug. Do any of the major OEMs sell any Phenoms by the way? I don't think I've seen any Dells or HPs with Phenom yet either.
OEMs just face the big problem that they can't sell systems with the fix disabled because of potential stability problems... and they can't sell them with the fix enabled because the performance wouldn't be good value for money. The Spider platform is not very attractive to OEMs at this point.

I don't know why tweakers.net is one of the few sites that actually dares to mention the truth... Perhaps other sites 'depend' on AMD too much.
So only one site tells the truth? And the article even starts with mentioning about Intel :D
Like you said, if you haven't seen it, doesn't mean that it's not there.
http://www.hardforum.com/showthread.php?t=1258024
 
So only one site tells the truth? And the article even starts with mentioning about Intel :D

I just posted the article where I read it.
There are probably other articles out there, but I don't feel the need to waste my time finding them. I know Wouter Tinus pretty well, I know he has connections in the industry, and he knows what he's talking about.

Also I don't see why mentioning Intel is bad... They are referring to Henri Richard (AMD marketing director)... see the picture? It's Henri Richard who is comparing Barcelona to Clovertown, and claiming Barcelona is going to 'blow away' Clovertown... lol
 
No, none of the C2D errata are as severe as the Phenom one. [. . .] that doesn't make it as unstable as an unpatched Phenom system.
Scali2 said:
In a way it's misinformation, or rather misleading information, because it's a very vague statement. It's nothing more than an opinion, not based on any quantitative analysis or fact. Words like 'rare', 'most users', 'much of a concern' really don't say anything, and leave it all up to the reader's interpretation.
You contradict yourself at every single turn. It really is ridiculous, Scali.

You say things like "No, none of the C2D errata are as severe as the Phenom one" and "[the C2D is not] as unstable as an unpatched Phenom system," and then turn around and bash others for using fuzzy language. How much more severe is the outcome of hitting the erratum on Phenom vs. C2D? How much more often does the Phenom run into errata than C2D? 1.1x as bad? 1.5x? 2.0x? You have no clue. You haven't done and do not have access to "any quantitative analysis or fact," as you put it. You're leaving the severity of the error up to the reader and trying to spin it as bad for AMD as you can possibly get away with. I don't see how everyone here is letting you get away with bashing people for doing the exact same thing you yourself are doing.

Sorry if this is a little inflammatory, but you've really topped yourself here in this thread as far as FUDing is concerned; someone had to call you out on it. Flame on if you feel like it; you're on ignore, and I won't even see your response. You can waste your time writing one if you feel like it. The sad thing is that there's nothing else you can say to make up for your past hypocrisy and FUDing. The posts you've made in the past are too old to edit or remove. They will always be there, and they show your true intent and personal character better than anything I could ever say about you.
 
You contradict yourself at every single turn. It really is ridiculous, Scali.

You say things like "No, none of the C2D errata are as severe as the Phenom one" and "[the C2D is not] as unstable as an unpatched Phenom system," and then turn around and bash others for using fuzzy language. How much more severe is the outcome of hitting the erratum on Phenom vs. C2D? How much more often does the Phenom run into errata than C2D? 1.1x as bad? 1.5x? 2.0x? You have no clue. You haven't done, and do not have access to "any quantitative analysis or fact," as you put it. You're leaving the severity of the error up to the reader and trying to spin it as bad for AMD as you can possibly get away with. I don't see how everyone here is letting you get away with bashing people for doing the exact same thing you yourself are doing.

Nice try, bud, however, the Intel errata are purely cosmetical... Even Linus Torvalds agrees on that (he accused Theo de Raadt of making a big deal out of nothing).
However, there is a Linux workaround available for the Phenom bug.
The Phenom bug has been reproduced in live environments.
The Core2 bugs are all purely theoretical, and cannot be reproduced with any commercial software.
Ofcourse you could try accusing Linus Torvalds of spreading FUD. In fact, why not accuse him of being a fanboy? Because he stood up and defended Intel for their bugs, but with AMD he keeps his silence.

Bottom line is: the K10 bug is real, and in that case it is important to know how likely you are going to run into it.
The Core2 Duo bugs don't affect real-life environments at all... They only occur in theoretical situations, not in practical ones.
Proof is simple: After rigourous testing, no OS has had to have kernel modifications for workarounds (the errata are known and well-documented), and no OEM has encountered problems during or after validation, no OEM stopped shipping Core2 Duos at any point in time, not even for servers, and there has been no report of any Core2 Duo bugs spotted in the wild so far.
Even if a bug on C2D would ever pop up in the wild now, K10 already has a healthly headstart on bug occurences, so C2D is never going to catch up to K10's level of severity anymore. QED

I believe the particular microcode that was updated, was related to a part of the TLB that was previously undocumented (I believe it had to do with whether a certain state would be restored automatically after a certain event, or if the OS had to reset the state manually, to make sure). For C2D, Intel documented this part of the TLB... However, some OS developers found out that the undocumented behaviour was different on older CPUs, so software that would make use of it (none of which currently exists in regular commercial software), would have problems on one or the other. So Intel issued the microcode update to make sure all CPUs have the same, documented behaviour from now on.
Perfect example of a cosmetic fix.
 
Back
Top