Phenom VS. Intel Quadcores (Q6600) ?

IBTL +1

I recently upgraded from a 939 x2 3800 to a Q6600, in my situation it made good sence as i had to purchase a new systemboard, ram, and CPU so I took the path that at this point has more longevity. At the time of my upgrade the Phenom was about 200$ compared to the Q6600 oem at 260$ so for 60$ more I was able to put together a system that should have a few years of good productivity in it.

What GPU did you get?
 
I got a new (to me) ati 1950gt pcie from flee bay. I may upgrade in the future possibly around the holidays again. At this point I am mainly running VM, and the occasional game. My LCD's max at 1280x1024 (I got 2 HP LP1965's) so anything above this resolution at this time is unnecessary.
 
This thread has outlived its usefulness. What people are arguing over here isn't worth being argued over.

I think we can all agree on the following:

1) Intel currently has the performance crown.
2) In general, Intel quad-cores will output as much as or less heat than their comparable Phenom counterparts.
3) Phenom is slightly cheaper. This may appeal to some, especially those who already have a compatible AM2 motherboard.
4) Phenom currently has a potentially serious bug concerning address translation.
a) You can currently garuantee stability by turning on a BIOS fix on most motherboards. This fix will incur about a 10% performance hit.
b) You can turn off the fix and have full performance, and in all likelyhood you will never experience the effects of the issue. However, nobody can currently garuantee that you will not experience a system hang, crash, or worse, silent failure leading to data corruption, etc, at some point. The words of AMD's own employees state that the issue can be experienced outside of virtualization workloads.
c) Users who buy an Intel quad-core will not have to choose between potential instability and a performance hit.
d) If you still want a Phenom, but don't want to choose between a performance hit and potential instability, you can wait several months and buy a Phenom with the TLB issue fixed.​

If we can all accept the above, what is there to argue about, seriously? Arguing about the severity of the TLB issue is like an argument between people who error out on Prime95 but don't experience any other issues, and others who won't use an overclock 24/7 unless it's Prime stable for X period of time. Some people are willing to use their computer in conditions that could potentially cause an error, and others are much less tolerant of even potential failure. Is the person willing to run the risk an idiot if they never experience any issues outside of Prime? Is the more cautious overclocker an idiot for not wanting to make the choice between performance and stability? That may not be the best example, but the arguments are still valid over this whole errata thing.

Are we seriously going to start crucifying people here for not buying Phenoms? Or for even offering the truth and other options to inquisitive buyers?
 
I got a new (to me) ati 1950gt pcie from flee bay. I may upgrade in the future possibly around the holidays again. At this point I am mainly running VM, and the occasional game. My LCD's max at 1280x1024 (I got 2 HP LP1965's) so anything above this resolution at this time is unnecessary.
In your situation, I would recommend the Q6600 because you need more CPU processing power. However for others who thing that they will get more gaming performance with the Q6600, I would say that with the $60 difference you can get a better GPU than the X1950GT and you can get a better performance in games with a HD3850 instead of a X1950GT (considering the new price) even with a slower Phenom.


This thread has outlived its usefulness. What people are arguing over here isn't worth being argued over.

I think we can all agree on the following:

1) Intel currently has the performance crown.
2) In general, Intel quad-cores will output as much as or less heat than their comparable Phenom counterparts.
3) Phenom is slightly cheaper. This may appeal to some, especially those who already have a compatible AM2 motherboard.
4) Phenom currently has a potentially serious bug concerning address translation.
a) You can currently garuantee stability by turning on a BIOS fix on most motherboards. This fix will incur about a 10% performance hit.
b) You can turn off the fix and have full performance, and in all likelyhood you will never experience the effects of the issue. However, nobody can currently garuantee that you will not experience a system hang, crash, or worse, silent failure leading to data corruption, etc, at some point. The words of AMD's own employees state that the issue can be experienced outside of virtualization workloads.
c) Users who buy an Intel quad-core will not have to choose between potential instability and a performance hit.
d) If you still want a Phenom, but don't want to choose between a performance hit and potential instability, you can wait several months and buy a Phenom with the TLB issue fixed.​

If we can all accept the above, what is there to argue about, seriously? Arguing about the severity of the TLB issue is like an argument between people who error out on Prime95 but don't experience any other issues, and others who won't use an overclock 24/7 unless it's Prime stable for X period of time. Some people are willing to use their computer in conditions that could potentially cause an error, and others are much less tolerant of even potential failure. Is the person willing to run the risk an idiot if they never experience any issues outside of Prime? Is the more cautious overclocker an idiot for not wanting to make the choice between performance and stability? That may not be the best example, but the arguments are still valid over this whole errata thing.

Are we seriously going to start crucifying people here for not buying Phenoms? Or for even offering the truth and other options to inquisitive buyers?

1) Agree

2) Let see again the numbers:
Q6600 100% Load: 160W
Q6600 75% Load: 156W ======9600 100% Load: 155W
Q6600 50% Load: 150W ======9600 75% Load: 151W
Q6600 25% Load: 137W ======9600 50% Load: 140W
========================9600 25% Load: 128W
Q6600 0% Load: 108W =======9600 0% Load: 105W

3) Agree but I need to add that you will get less performance with the cheaper CPU

4) It has the bug but it is not as serious like you said, and your analogy is way off, it's not like you have errors on Prime95 but it's more like when you have a really huge OC, you passed Prime95 for several hours but once in a few weeks or months, you will get a crash when doing something really CPU intensive. If you read carefully this thread and the XtremeSystems thread linked from this thread, you will know how hard it is to reproduce the bug. Even with several hours of Prime95, you will not get any errors and any crash. Like I said before, I have built two Phenom systems with the bug fix turned off and never had any crash in weeks, Manny also said the same thing so why would you say that it is like having a CPU with Prime95 errors? Unless you have a Phenom system and experience the bug yourself or know someone who has a Phenom system and has experienced the bug, you are making this issue bigger than it really is.
 
Over 15% having problems is not important?
If it was under 1% you'd have a point, but 15% is just way too much to be talking about 'extremely rare' or 'most users will probably never see the bug'.

Yet the 15% who voted yes said nothing in the thread. Anyone who loves Intel or hates AMD can vote yes and of course the other way around too.
 

He said *comparable* counterparts. The 9600 is not comparable, it's slower.
The comparable part is the 9900, performance-wise.
But look, it uses more power.
Same as normalizing these 9600 results to the performance... Q6600 will still win out on performance-per-watt.
A Q6600 will either finish the job quicker, so it will run at 100% for a shorter period of time than a 9600, or it will be running at a lower CPU load than the 9600... You have to take that into consideration aswell.

4) It has the bug but it is not as serious like you said, and your analogy is way off, it's not like you have errors on Prime95 but it's more like when you have a really huge OC, you passed Prime95 for several hours but once in a few weeks or months, you will get a crash when doing something really CPU intensive. If you read carefully this thread and the XtremeSystems thread linked from this thread, you will know how hard it is to reproduce the bug. Even with several hours of Prime95, you will not get any errors and any crash. Like I said before, I have built two Phenom systems with the bug fix turned off and never had any crash in weeks, Manny also said the same thing so why would you say that it is like having a CPU with Prime95 errors? Unless you have a Phenom system and experience the bug yourself or know someone who has a Phenom system and has experienced the bug, you are making this issue bigger than it really is.

Prime95 is a stresstest that works well for testing the stability of an overclocked system.
The equivalent of this for the TLB bug is to run virtualization and a 100% system load for any length of time. Why don't you and Manny run that for a few days, and see if the system crashes?
Then we will at least have a rough idea about the worst case scenario for this bug.
 
He said *comparable* counterparts. The 9600 is not comparable, it's slower.
The comparable part is the 9900, performance-wise.
But look, it uses more power.
Same as normalizing these 9600 results to the performance... Q6600 will still win out on performance-per-watt.
A Q6600 will either finish the job quicker, so it will run at 100% for a shorter period of time than a 9600, or it will be running at a lower CPU load than the 9600... You have to take that into consideration aswell..
9600 is the comparable part price-wise, clock speed-wise and performance-wise. 9900 doesn't exist yet and we don't know how bad the power consumption of the B3 Phenom.

Prime95 is a stresstest that works well for testing the stability of an overclocked system.
The equivalent of this for the TLB bug is to run virtualization and a 100% system load for any length of time. Why don't you and Manny run that for a few days, and see if the system crashes?
Then we will at least have a rough idea about the worst case scenario for this bug.
Try that first with an OCed Q6600 without Prime95 errors but OCed to the limit (>3.5GHz). Can you guarantee that you won't get a crash? At least I can guarantee that I won't get a crash in that situation with the Phenom by turning on the fix. Saying using a Phenom without the fix is like using an OCed chip with Prime95 errors is way too exagerated.
 
Try that first with an OCed Q6600 without Prime95 errors but OCed to the limit (>3.5GHz). Can you guarantee that you won't get a crash? At least I can guarantee that I won't get a crash in that situation with the Phenom by turning on the fix. Saying using a Phenom without the fix is like using an OCed chip with Prime95 errors is way too exagerated.

I think he's trying to draw a parallel between the two; whether it's exaggerated or not, I think it still proves a point.

In any event, yes, you can guarantee that a Q6600 (or any other processor for that matter) won't error out of Prime95 any more than at stock settings if you've got the right cooling/settings. In other words, you can guarantee it if you have stable settings.

I think this whole argument over the TLB erratum has gone on for way too long. It will be moot in a month or so anyway, so lets just leave it at that.
 
9600 is the comparable part price-wise, clock speed-wise and performance-wise. 9900 doesn't exist yet and we don't know how bad the power consumption of the B3 Phenom.

No, 9600 is cheaper, and has lower performance.
It also has lower clockspeed, but I don't see why you want to bring clockspeed into this. You haven't learnt from the whole MHz-myth days with P3, P4 and Athlon?
9900 may not be on sale yet, but it has been included in various reviews, and it came closest to the Q6000 in performance.
See here for example: http://www.xbitlabs.com/articles/cpu/display/amd-phenom_10.html#sect1
However, it had far higher power consumption: http://www.xbitlabs.com/articles/cpu/display/amd-phenom_15.html#sect0

I don't think a new stepping will close a gap of about 50W, do you?
Looking at historical data, 20W might be expected... 30W could happen if AMD gets exceptionally lucky, but 50W would be nothing short of a miracle.

Try that first with an OCed Q6600 without Prime95 errors but OCed to the limit (>3.5GHz). Can you guarantee that you won't get a crash? At least I can guarantee that I won't get a crash in that situation with the Phenom by turning on the fix. Saying using a Phenom without the fix is like using an OCed chip with Prime95 errors is way too exagerated.

Oh please. That just didn't make any sense whatsoever.
I'm just saying that you should run a stresstest so we can see how often the bug actually occurs.
Since you're the one claiming that the TLB is such a non-issue that you may never actually encounter it, I'd like to see how long it takes for the bug to occur in a stress-test. Then we know that it will at least take as long as that with regular use, probably much longer.
This is a very important issue, because running with or without TLB fix means 10-20% performance difference, which directly affects the value-for-money of the CPU.

I'd also like to remark that you'd still be running the CPU within official AMD specs, where it *should* just be working properly.
Overclocking a CPU is just running it outside the specs it was designed for, and all guarantees about performance, reliability etc (and your warranty) are void, as far as the manufacturer is concerned.
 
I think he's trying to draw a parallel between the two; whether it's exaggerated or not, I think it still proves a point.

I'm just saying that Prime95 more or less answers the question "How stable is your system?".
You can just answer it by "X hours Prime95".
Since the TLB issue requires a different stresstest (while Prime95 stresses various critical components of the CPU, it's not the TLB that's being stressed), I'd like to know what X is on an average Phenom system with no TLB fix installed (since alg7_munif is recommending people here to run without the fix).
So I'd like to see some hard numbers on this.
 
Oh please. That just didn't make any sense whatsoever.
I'm just saying that you should run a stresstest so we can see how often the bug actually occurs.
Since you're the one claiming that the TLB is such a non-issue that you may never actually encounter it, I'd like to see how long it takes for the bug to occur in a stress-test. Then we know that it will at least take as long as that with regular use, probably much longer.
This is a very important issue, because running with or without TLB fix means 10-20% performance difference, which directly affects the value-for-money of the CPU.

I'd also like to remark that you'd still be running the CPU within official AMD specs, where it *should* just be working properly.
Overclocking a CPU is just running it outside the specs it was designed for, and all guarantees about performance, reliability etc (and your warranty) are void, as far as the manufacturer is concerned.
Other people and me have run Prime95, AOD Stress test and other stress test without crashing for hours without the fix. Btw, running the chip without the fix is out of AMD spec, AMD spec required the fix enabled by default, it is up to the user themselves to chosse whether they want to disable it or OC their chip. I recommend people to disable the fix just like a lot of people recommended others to OC their Q6600. So when you OC your chip, you would run Prime95 but not virtualization for stress testing, however when you disable the fix, Scali2 suggests that running Prime95 alone is not enough? :rolleyes: Doesn't make anysemse to me.
Seriously, I think the TLB issue is blown way out of proportion. Just look at these threads:

http://www.xtremesystems.org/forums/showthread.php?t=170462
http://www.xtremesystems.org/forums/showthread.php?t=172026

Even with all 4 cores at 100% load with certain benchmarks, its very hard to reproduce the bugs.
I agree with quadnad, this argument over the TLB erratum has gone on for way too long, I would not argue it anymore and I would just take the the words of those who have actually used the CPU themselves over anyone who is known to be BIASED to a certain brand. Anyway I wonder how many of those OCed Q6600 users can guarantee that their CPU will pass Scali2's virtualization stress test and do you really care that your OCed Q6600 would pass Scali2's virtualization stress test or not?
 
Many others and me have run Prime95, AOD Stress test and other stress test without crashing for hours without the fix.

But you're not stress-testing the TLB!
What you're saying is like with the Pentium FPU bug.... "I've been running integer stress tests all day, and haven't seen the FPU bug occur"... ofcourse not, you're not stressing the FPU.

You don't seem to get that I'm not arguing with you about the severity of the bug...
Thing is just that you constantly want to stress how rare the bug is, but you never give any quantification.
I'm arguing that you fail to specify the exact severity of the bug.
So put a number on it: In the worst case scenario, how long does it take on average for the bug to occur? (we'd like to see multiple CPUs, say about 25... and multiple runs for any kind of accuracy).
Then we know what we're talking about.
 
But you're not stress-testing the TLB!
What you're saying is like with the Pentium FPU bug.... "I've been running integer stress tests all day, and haven't seen the FPU bug occur"... ofcourse not, you're not stressing the FPU.

You don't seem to get that I'm not arguing with you about the severity of the bug...
Thing is just that you constantly want to stress how rare the bug is, but you never give any quantification.
I'm arguing that you fail to specify the exact severity of the bug.
So put a number on it: In the worst case scenario, how long does it take on average for the bug to occur? (we'd like to see multiple CPUs, say about 25... and multiple runs for any kind of accuracy).
Then we know what we're talking about.

Do you want to know how exactly unstable my OCed Q6600 in virtualization too? I never had any crash with Prime95 with my OCed Q6600 just like my friend never had any crash with his Phenom, but I guess that I never give any quantification on how stable my OCed Q6600 is.
 
Do you want to know how exactly unstable my OCed Q6600 in virtualization too?

What does that have to do with anything?
Why are you avoiding the question?

I'm just pointing out that nobody seems to realize what a TLB does and how it is to be stressed.
Running Prime or Julia benchmarks? That's not what they mean by a 'heavy load'!
You have to have many context switches, and preferably lots of cache/memory usage, so the TLB has to get updated a load.
Running a single process with a very limited memory footprint isn't going to bother the TLB at all, even though your CPU may be running at 100% on all cores.

I suppose you also realize that you can not prove that the bug will NEVER occur, because that means you will have to continue testing into eternity? It's not scientific.
You may want to read up on some of Popper's work: http://en.wikipedia.org/wiki/Falsifiability
 
Do you want to know how exactly unstable my OCed Q6600 in virtualization too?

Again, to be fair, if your overclock is unstable then that's the fault of something in your setup. Whether your chip just won't do 3.2 stable is one thing, or (more likely) it could be a combination of a variety of settings, or it could be a heat issue. If done properly, overclocked chips should be just as stable as at stock.
 
Again, to be fair, if your overclock is unstable then that's the fault of something in your setup. Whether your chip just won't do 3.2 stable is one thing, or (more likely) it could be a combination of a variety of settings, or it could be a heat issue. If done properly, overclocked chips should be just as stable as at stock.

My OC is stable in Prime95 just like the Phenom is stable in Prime95. However I can't guarantee that my OC is stable in Scali2's virtualization stress test just like I can't guarantee the how long the Phenom will be stable in Scali2's virtualization stress test without the fix and I don't have a plan to test whether my OCed Q6600 is stable in Scali2's virtualization stress test.

I can however guarantee that my Q6600 will be stable in Scali2's virtualization stress test at stock speed and the Phenom will be stable in Scali2's virtualization stress test with the fix enabled.

Like I said before that running the Phenom without the fix is exactly like running an OCed Q6600. A lot of people have run their Phenom without the fix and they have a stable system in Prime95 and other stress test but not in Scali2's virtualization stress test just like a lot of people run their Q6600 OCed and said that their OCed Q6600 is stable in Prime95 and other stress test but how many of those OCed Q6600 user can claim that their Q6600 is stable in Scali2's virtualization stress test?
 
My OC is stable in Prime95 just like the Phenom is stable in Prime95. I can't guarantee that my OC will be stable in Scali2's virtualization stress test just like I can't guarantee the how long the Phenom will be stable without the fix. I can however guarantee that my Q6600 will be stable in Scali2's virtualization stress test at stock speed and the Phenom will be stable with the fix enabled.

We all know it's not going to be stable... What we want to know is: HOW unstable is it?
You yourself started all this crap about how rare the bug is. Put some numbers on it.
After all, you're advising people to buy a Phenom because you can run it perfectly fine without the fix installed.
Even you would have to agree that the Phenom with the fix installed is a bad buy, even at the low price, because the performance is just appalling, and power consumption goes up even further because of the fix.
So put some numbers on it... What are we looking at when we are running an average Phenom system without the fix, worst case scenario?
 
We all know it's not going to be stable... What we want to know is: HOW unstable is it?
You yourself started all this crap about how rare the bug is. Put some numbers on it.
After all, you're advising people to buy a Phenom because you can run it perfectly fine without the fix installed.
Even you would have to agree that the Phenom with the fix installed is a bad buy, even at the low price, because the performance is just appalling, and power consumption goes up even further because of the fix.
So put some numbers on it... What are we looking at when we are running an average Phenom system without the fix, worst case scenario?

How rare? So rare that it is stable in Prime95 and AOD stress test for hours, the system also never crashed for weeks now, so is that not enough numbers to show how rare it is?

It is not as bad as running an OCed CPU with Prime95 errors.

How unstable the Phenom in Scali2's virtualization stress test? Would you like to know how unstable my OCed Q6600 in in Scali2's virtualization stress test as well?
 
How rare? So rare that it is stable in Prime95 and AOD stress test for hours, the system also never crashed for weeks now, so is that not enough numbers to show how rare it is?

It is not as bad as running an OCed CPU with Prime95 errors.

How unstable the Phenom in Scali2's virtualization stress test? Would you like to know how unstable my OCed Q6600 in in Scali2's virtualization stress test as well?

You don't get it, do you?
 
You don't get it, do you?

No, you don't get it, running a Phenom without a fix is just like running a Q6600 at OCed speed, no one can guarantee that their OCed CPU is stable in each and every stress test including but not limited to Scali2's virtualization stress test.

What Intel and AMD guaranteed is that their CPU should be working as advertised at stock speed and settings which include but no limited to stock voltage, stock clock speed, stock FSB and in AMD's case, the TLB bug fix is enabled.

Running the CPU out of this spec is a risk taken by the user themselves. Like I said before, no one can guarantee that their OCed CPU is stable in each and every stress test including but not limited to Scali2's virtualization stress test. However compairing running a Phenom without the fix with running an OCed CPU with Prime95 errors is way too exaggerating.

Running Prime95 without errors for several hours is enough for some to claim that their OC is stable but without further testing, even they can't guarantee that their OC will pass other stress test including but not limited toScali2's virtualization stress test.

It would be better if they could do other stress testing including but not limited to Scali2's virtualization stress test but is it that critical to go to that extent just to satisfy someone who is known to be BIASED towards a certain brand in a forum when a Prime95 stress test is already considered sufficient stable by most user and the OCer himself especially when most user and himself have never encountered any crash due to the OC for weeks or months?

With that said, I think that the issue is way exaggerated than what it actually is. You can listen to the negative assumptions if you want to or you could just accept what the actual Phenom user have to say. Maybe I'm biased so do take my words for it but if you look at Manny's post history, I don't see any reason to doubt his experience or Kyle's.
 
What Intel and AMD guaranteed is that their CPU should be working as advertised at stock speed and settings which include but no limited to stock voltage, stock clock speed, stock FSB and in AMD's case, the TLB bug fix is enabled.

Okay, so we shouldn't disable the TLB fix then... Why did you recommend it anyway?

However compairing running a Phenom without the fix with running an OCed CPU with Prime95 errors is way too exaggerating.

Is it? I haven't seen any data on the subject yet. That's what I've been asking for, and you've been avoiding all this time.

Running Prime95 without errors for several hours is enough for some to claim that their OC is stable but without further testing, even they can't guarantee that their OC will pass other stress test including but not limited toScali2's virtualization stress test.

Yes, it is, for CPUs not troubled by a certain stability bug. The TLB bug changes the rules on what is to be considered stable, and therefore different stress tests are in order.
You don't need anything other than Prime95 to test the stability of a Q6600 because there are no stability issues. The only things that affect stability are the usual factors of voltage, heat, CPU load and all that.
Phenom adds TLB issues to the mix, and Prime95 does not address these issues.

It would be better if they could do other stress testing including but not limited to Scali2's virtualization stress test but is it that critical to go to that extent just to satisfy someone who is known to be BIASED towards a certain brand in a forum when a Prime95 stress test is already considered sufficient stable by most user and the OCer himself especially when most user and himself have never encountered any crash due to the OC for weeks or months?

Firstly, you're not just satisfying me, because there are tons of people on tons of forums asking just how stable or unstable the Phenom is. Why is nobody producing any data on this? What do you have to hide?
Secondly I am not biased to a certain brand, and I don't see why you have to throw that into your 'argumentation'. I will report you, because I am insulted by this.
You are probably the only one who considers me biased, now that duby is gone. Hey, what a coincidence, duby was so extremely out there in the AMD camp that he couldn't tell a neutral person from an Intel fanboy.
I am perfectly neutral and objective, and I pride myself in that. Anyone who thinks otherwise is probably severely biased themselves.
People who know how much I hate x86 in general, both Intel and AMD, will know how laughable it is to think I could ever be a fan of either company. Insert "Amiga Rulez" here ;)

With that said, I think that the issue is way exaggerated than what it actually is.

Why do you think that? Have any numbers on it? What are we talking about here? I have no idea what you consider 'exaggerated' or 'normal'. Numbers please.

Maybe I'm biased so do take my words for it but if you look at Manny's post history, I don't see any reason to doubt his experience or Kyle's.

Neither has done any stresstesting on the TLB.
No offense, but I'm not sure if either really understands the deeper meaning and consequences of this bug.
 
Okay, so we shouldn't disable the TLB fix then... Why did you recommend it anyway?
Should we OC our CPU? Why people recommend it anyway?
Is it? I haven't seen any data on the subject yet. That's what I've been asking for, and you've been avoiding all this time.
Prime95 would not give any errors with the fix disabled, it that not enough data already?
Yes, it is, for CPUs not troubled by a certain stability bug. The TLB bug changes the rules on what is to be considered stable, and therefore different stress tests are in order.
You don't need anything other than Prime95 to test the stability of a Q6600 because there are no stability issues. The only things that affect stability are the usual factors of voltage, heat, CPU load and all that.
Phenom adds TLB issues to the mix, and Prime95 does not address these issues.
Can you guarantee that any OCed Q6600 that has passed the Prime95 stability testing will also pass each and every other stress test including but not limited to Scali2's virtualization stress test?

I think that we're going in circles so I don't think that I need to answer this anymore unless some else really thinks the data provided is not enough.
 
Should we OC our CPU? Why people recommend it anyway?

Overclocking is a pretty 'analog' thing. You can choose how far you want to overclock it. You know you're starting from a stable setup at stock speeds, then you can just increase and find the 'sweet spot' in terms of performance and stability.
This is all well-understood, and each overclocker can decide for himself where that sweet spot is exactly.

TLB fix is just binary. Turn it on, and you're stable... Turn it off and you land somewhere in the instability zone... But where do you land exactly?
Nobody knows, and you're just blowing a lot of smoke.

Prime95 would not give any errors with the fix disabled, it that not enough data already?

No it isn't. If you actually understood what the bug is about, you'd realize this too. Come on, just be man about it and admit that you don't know what the bug means.
Or prove that you do, by explaining it to us.

Can you guarantee that any OCed Q6600 that has passed the Prime95 stability testing will also pass each and every other stress test including but not limited to Scali2's virtualization stress test?

No, but I can guarantee that you can get all stress tests stable if you back off the OC a bit, on a processor with no critical stability bugs like the Q6600. See the difference?
 
No, but I can guarantee that you can get all stress tests stable if you back off the OC a bit, on a processor with no critical stability bugs like the Q6600. See the difference?

And you will get less performance, just like when you turn on the fix, see the similiarity?
 
If you could sum up your point into a few sentences, what would they be?
I said it before but let me shorten it up for you:
Running a Phenom without a fix is just like running a Q6600 at OCed speed, no one can guarantee that their OCed CPU is stable in each and every stress test.

What Intel and AMD guaranteed is that their CPU should be working as advertised at stock speed and settings which include but no limited to stock voltage, stock clock speed, stock FSB and in AMD's case, the TLB bug fix is enabled.

Running the CPU out of this spec is a risk taken by the user themselves. Like I said before, no one can guarantee that their OCed CPU is stable in each and every stress test. However compairing running a Phenom without the fix with running an OCed CPU with Prime95 errors is way too exaggerating.

Running Prime95 without errors for several hours is enough for some to claim that their OC is stable but without further testing, even they can't guarantee that their OC will pass other stress test.

It would be better if they could do other stress testing including but not limited to Scali2's virtualization stress test but is it that critical to go to that extent when a Prime95 stress test is already considered sufficient stable by most user and the OCer himself especially when most user and himself have never encountered any crash due to the OC(or turning off the fix) for weeks or months?

With that said, I think that the issue is way exaggerated than what it actually is. You can listen to the negative assumptions if you want to or you could just accept what the actual Phenom user have to say. Maybe I'm biased so do take my words for it but if you look at Manny's post history, I don't see any reason to doubt his experience or Kyle's.
 
Neither has done any stresstesting on the TLB.
No offense, but I'm not sure if either really understands the deeper meaning and consequences of this bug.




No,I have not stress tested for the TLB errata,but the people I built Phenom systems for do not have any need for Virtualization software.I didnt think I had to test for it,and every other possible thing that might arise,given these were built for close friends who really
like AMD,and wanted a faster gaming PC then what they had at the time.

I did my job,and feel no guilt or worry over not testing for some beyond incredibly obscure
'bug'


Honestly Scali,I think you and a few others are making a bit too much of it.All this rumour and innuendo about it all over the net,has in my eyes hurt AMD financially.Not telling you or anyone else to hush up,(quite the contrary actually),only that its being blown so far out of proportion its insane !

I was informed via email that my BE is on its way to me now as of this morning.

I am hoping for 2.90 to 3Ghz,but dont expect anymore then 2.8Ghz with the high end air cooling I have access to right now.

TBH I am far,far more concerned with getting my grubby little paws on a 8900 GX2 ASAP,
.then building this up as soon as it gets here.I am aware based on past expeirince with Phenom builds,that this endevor will likely be a long and arduos one as to the oc itself.

I have tremendous patience with PC hardware,but even that only goes so far. :)

Anyway I am off to classes.
 
I own 2 Q6600 G-0s and one Agena 9600. The 9600 will do about 2.5 and some change stable. Both Q6600s will do 3.0GHz on the stock cooler @ stock voltage effortlessly. 3.2 only requires 1.36v and they can Prime for days. One goes up to 3.8GHz on air and the other 3.7GHz.

If you overclock and can only afford to build a single rig go Q6600 G-0 or wait for the Q9450.

BTW the voltage requirements and heat output ramp up considerably for phenom @ 2.6 and beyond.
 
No,I have not stress tested for the TLB errata,but the people I built Phenom systems for do not have any need for Virtualization software.I didnt think I had to test for it,and every other possible thing that might arise,given these were built for close friends who really
like AMD,and wanted a faster gaming PC then what they had at the time.

I did my job,and feel no guilt or worry over not testing for some beyond incredibly obscure
'bug'


Honestly Scali,I think you and a few others are making a bit too much of it.All this rumour and innuendo about it all over the net,has in my eyes hurt AMD financially.Not telling you or anyone else to hush up,(quite the contrary actually),only that its being blown so far out of proportion its insane !

Why are you getting defensive?
I'm not making anything out of this bug. Nowhere have I claimed anything about how often it occurs or how severely you will be affected.
In fact, I'm here to try and find out exactly that, because there seems to be no data on this subject.
Why does everyone with a Phenom get so defensive (and in effect are making a big deal out of nothing... just test the thing and get it over with)?
Must be the psychological aspect to the bug. But I'm not interested in psychology, I'd just like a scientific approach to determining how rare this bug really is.
I'm not trying to hurt AMD. In fact, it may actually help AMD if the results are positive enough.
 
Why are you getting defensive?
I'm not making anything out of this bug. Nowhere have I claimed anything about how often it occurs or how severely you will be affected.
In fact, I'm here to try and find out exactly that, because there seems to be no data on this subject.
Why does everyone with a Phenom get so defensive (and in effect are making a big deal out of nothing... just test the thing and get it over with)?
Must be the psychological aspect to the bug. But I'm not interested in psychology, I'd just like a scientific approach to determining how rare this bug really is.
I'm not trying to hurt AMD. In fact, it may actually help AMD if the results are positive enough.


I think "marketing aspect to the bug" would be much more appropriate. Or better yet, "propaganda aspect". ;)
 
Back
Top