Intel's Pat Gelsinger Exclaims "Intel is Back" AMD is "Over"

[GURU7OF9]

Fingers crossed for AMDs driver squad, but to be fair it's going to be On AMD , windows 11 can't have slipped under they're radar, I'm not keen on unfair comparison but the truth tends to wash out doesn't it.
What can anyone but AMD do, the opposite (W10 benches)is also unfair.

Which gets back to testing both ADL and Ryzen on both Win10 and Win 11 .
You will get a broader fairer spread of results.
You would expect it to show ADL is faster on Win11 and Ryzen is faster on Win 10 at this point in time and as for comparisons against one one another we shall see in due course.
Even if Ryzen was fixed now for Win11, its probly too late to be included !
I honestly can't see any other way to do it fairly but we already know that wont happen here ! Obviously it would be double the work !
Be interesting to see what other reviewers choose to do !

 

[zlobby]

Hey @W1zzard, sorry for asking but are you planning to take a look at the root cause of the cache issues? I'm sure you and your crew are pretty much up to task.
To be honest, so am I but I'm too lazy and I don't care about the implications. I'm just curious about the engineering aspects.

 

[Valantar]

https://twitter.com/i/web/status/1448310872812650510

From people responding on Twitter, it seems ADL is faster in the CPU render score but slower in the CPU simulation score. Overall seems to very slightly favor ADL. Still (and with all the caveats of this being an unsubstantiated leak, a single test, etc.) that looks like a tie to me.

 

[outpt]

Intel where back. Back in the janitors closet at microstiffe checking each other’s schedulers. I’ll be over at your closet tomorrow.

 

[Richards]

I'm all for interesting investigation.

You are implying no one would use win 10 on Aderlake, for gaming that makes sense, but actually for office, work based tests the opposite is true since most companies will take years to adopt windows 11, just as they took years moving off every other OS, shit test Xp too, no I jest

And regardless Intel , Microsoft and AMD have certainly created the perfect, benching shitfest for some poor soul, I wouldn't want to be that writer, he or she can't win here.

Windows 10 tanks alder lake performance and turns on the e cores only.. leaks from china are saying

 

[W1zzard]

Hey @W1zzard, sorry for asking but are you planning to take a look at the root cause of the cache issues? I'm sure you and your crew are pretty much up to task.
To be honest, so am I but I'm too lazy and I don't care about the implications. I'm just curious about the engineering aspects.

don't think this is worth spending a lot of time investigating, since a fix will be released and the issue will be solved forever

 

[zlobby]

don't think this is worth spending a lot of time investigating, since a fix will be released and the issue will be solved forever

But we'll never know the truth. Truth is absolute and eternal. But yeah, I'm too lazy to pursuit it.

Don't you have some wunderkind coder intern who could crack it in a few hours? Still nobody else got it out on the webs, so clicks and potential revenue is still for the taking.

 

[ratirt]

I'm quite sure the W11 issues will be fixed with AMD CPUs. I understand ADL is a new arch and prepared for the W11 with the Big.Little arch but as far as I understand AMD is supposedly going that route as well. (Not sure why though). I'm also pretty much confident that MS with the new W11 OS will not favor one arch vs the other because that is bad business for the company.

 

[GURU7OF9]

Fingers crossed for AMDs driver squad, but to be fair it's going to be On AMD , windows 11 can't have slipped under they're radar, I'm not keen on unfair comparison but the truth tends to wash out doesn't it.
What can anyone but AMD do, the opposite (W10 benches)is also unfair.

W10(Ryzen) verses W11(Intel) also isn't fair and I am not sure it would go as you might expect, I think W11 will have optimization to be done.

Why wouldnt Ryzen on win10 be fair "if its not broken" comparing it to ADL on win11 when ADL is specifically designed for it?

Seems a whole lot fairer to me than testing Ryzen on w11 when you know its broken!

If Ryzen is tested on w11 in broken state then that is just dumb and a totally biased review!

Thats just the same as not testing ADL on win10 cos it is broken and yet you rationalize that its ok the other way around!

Completely hypocritical !

 

[TheoneandonlyMrK]

Why wouldnt Ryzen on win10 be fair "if its not broken" comparing it to ADL on win11 when ADL is specifically designed for it?

Seems a whole lot fairer to me than testing Ryzen on w11 when you know its broken!

If Ryzen is tested on w11 in broken state then that is just dumb and a totally biased review!

Thats just the same as not testing ADL on win10 cos it is broken and yet you rationalize that its ok the other way around!

Completely hypocritical !

Because 11 adds adl which reduces performance on 11 not on 10 giving Ryzen on 10 an advantage.

 

[GURU7OF9]

How is 3

Because 11 adds adl which reduces performance on 11 not on 10 giving Ryzen on 10 an advantage.

How is ADL on w11 a disadvantage?
How is Ryzen on w11 in broken state not an advantage for ADL ?

 

[GURU7OF9]

No one seems to have taken any notice with ADL Z690 Motherboards, the only way you can run your super fast M.2 drive in PCIe 5.0 is to run the 16 lanes in 8x2 configuration.

Have to say thats a big let down in my book ! Extremely dissappointing for a brand new platform!

Look, it will most likely still work fine and have heaps of bandwidth for 8 lanes of 5.0 for both video card and m.2 drive, but seriously i cant beleive you have to gimp it to 8 x 2 lanes to make it work !

Dissappointing !

 

[lola46]

the I9 12900K has TDP 330W of jokes ,AMD Ryzen 9 5900X has 12 Cores and has TDP only 115W

 

[HenrySomeone]

Aha, sure it does "pal"

 

[londiste]

the I9 12900K has TDP 330W of jokes ,AMD Ryzen 9 5900X has 16 Cores and has TDP only 115W

This is the weird bit.
12900K does not consume 330W. Although incredibly inefficient at that, 5.2GHz OC is not something to scoff at. 12900K will probably be run with some stupid power limit at 200-something W from motherboards and run at 4.9GHz boost at that. When stock power limit is applied the base clock is 3.7GHz.
5900X has in most cases the power limit at ~140W which it can reach at full load.
And this is still primarily with some type of stress test or benchmark load that uses AVX2 and/or is otherwise evil.

In practice and normal usage the two will be closer together than headlines makes it seem.

 

[Valantar]

the I9 12900K has TDP 330W of jokes ,AMD Ryzen 9 5900X has 16 Cores and has TDP only 115W

I have no idea what you think TDP means, but whatever it is, it is not correct. TDP is a manufacturer-implemented specification for the cooling capacity for heatsinks. Power draw is not the same as TDP. Also, the 5900X has a 105W TDP and a 144W standard power limit. Recent Intel chips have had 125W TDPs with 225W (or thereabouts) PL1 short-term boost power limits. So whatever you're referring to, you're off by quite a bit. Overclocking also generally disables power limits.

 

[mtcn77]

I have no idea what you think TDP means, but whatever it is, it is not correct. TDP is a manufacturer-implemented specification for the cooling capacity for heatsinks. Power draw is not the same as TDP. Also, the 5900X has a 105W TDP and a 144W standard power limit. Recent Intel chips have had 125W TDPs with 225W (or thereabouts) PL1 short-term boost power limits. So whatever you're referring to, you're off by quite a bit. Overclocking also generally disables power limits.

It does not. There is no way you can disable it. Besides, tdp is very important since the heat conduction takes place at the IHS, yet the power density is too high for normal operation which coupled with ihs concavity makes it a lot more significant. Limiting edc is very important in ryzen series cpus since it avoids temperature gradient gap between the chip and the ihs when heat flow is steady, and not fluctuating.

 

[Valantar]

It does not. There is no way you can disable it. Besides, tdp is very important since the heat conduction takes place at the IHS, yet the power density is too high for normal operation which coupled with ihs concavity makes it a lot more significant. Limiting edc is very important in ryzen series cpus since it avoids temperature gradient gap between the chip and the ihs when heat flow is steady, and not fluctuating.

What on earth are you talking about? I said overclocking disables power limits. The rest of the post is specifically about why TDP and power limits of various kinds are not the same. Did you read the post at all, or what I was responding to?

 

[mtcn77]

What on earth are you talking about? I said overclocking disables power limits. The rest of the post is specifically about why TDP and power limits of various kinds are not the same. Did you read the post at all, or what I was responding to?

Don't flame. Overclocking does not delimit FID, end of story.

Re: Definitive guide to configuring the Ryzen 3900X

Ryzen already has a built in silicon fitness monitoring feature (FIT) which prevents voltages that AMD feels would reduce the expected life of the processor. "The Stilt" discusses it at length in his "Matisse, strictly technical piece" over on Overclock.net. Here are a couple excerpts. This...

community.amd.com

 

[Valantar]

Don't flame. Overclocking does not delimit FID, end of story.

Re: Definitive guide to configuring the Ryzen 3900X

Ryzen already has a built in silicon fitness monitoring feature (FIT) which prevents voltages that AMD feels would reduce the expected life of the processor. "The Stilt" discusses it at length in his "Matisse, strictly technical piece" over on Overclock.net. Here are a couple excerpts. This...

community.amd.com

I'm not flaming, I'm asking you what on earth you are saying, as your response makes zero sense in relation to my post. What is "FID"? What does "delimiting" it mean? And are you actually saying that overclocking - of which adjusting (or disabling, where possible) power limits is a fundamental component in modern architectures - doesn't override stock power limits? Or are you saying that TDP is equal to power draw? Because that was what I said in the post you responded to: TDP and power draw is not the same, and stock power limits are irrelevant when overclocked - specifically in relation to the leaks surrounding the (possible) 12900K and its rumored power draws when overclocked. Whatever extrapolation you have done beyond that is not my problem.

 

[GURU7OF9]

don't think this is worth spending a lot of time investigating, since a fix will be released and the issue will be solved forever

Looks like you are/were correct with the fix supposedly released now!
Have you got time to test or retest Ryzen on win 11 with the fixes in place?
Has any one done some basic tests to make sure the fixes are actually work properly?

Aha, sure it does "pal"

This posted graph of power is not just the cpu it is the WHOLE SYSTEM .
Even says it at top of chart under techpowerup heading.
My Ryzen 5600x does not run at 126w not even maxed out at pbo mb limits. It topped out at 121w. I dont bother with pbo I leave it at default which is 76W .

 

[lola46]

i heard in some article that even with AIO 360MM the intel still hits 92 degress Celcius
Intel Core i9-12900K Overclocked To 5.3 GHz Across All Cores With 1.44V Voltage, Hits Insane 400W Power Consumption In AIDA64 Stress Test (wccftech.com)

 

[Waldorf]

Lol, back to spending billions more than the other guy, while producing a product about 10% faster..

@Valantar
first you say its "..disables power limits," then it changes to "..adjusting limits, disabling when possible..". ok
changing P limits for oc and unlimited, are not the same thing,
and unless all (3 major) power settings can be run lets say at 2000A (disregarding cooling and such for a moment), its NOT "unlimited".

 

[Valantar]

@Valantar
first you say its "..disables power limits," then it changes to "..adjusting limits, disabling when possible..". ok
changing P limits for oc and unlimited, are not the same thing,
and unless all (3 major) power settings can be run lets say at 2000A (disregarding cooling and such for a moment), its NOT "unlimited".

It is effectively unlimited, as you can change them so that other factors (such as cooling) will always be the bottleneck. Also, when speaking of CPU power limits, the stock power limits are the only ones really worth talking about - as they are the only externally determined power limits - which are disabled as soon as any adjustment is made, no matter how small that adjustment is. Beyond that you're looking at a voluntary power limit, which ... well, it would be another thing entirely. When you can set the limit manually, it is no longer a limit, thus it is disabled. Pretty simple, really. Doesn't matter if you're unticking a box that says "power limits" or adjusting it to a higher number than can be cooled without exotic cooling; the effect is the same.

 

[GURU7OF9]

i heard in some article that even with AIO 360MM the intel still hits 92 degress Celcius
Intel Core i9-12900K Overclocked To 5.3 GHz Across All Cores With 1.44V Voltage, Hits Insane 400W Power Consumption In AIDA64 Stress Test (wccftech.com)

For 12900k to reach 400W should that be true, is kinda insanely bad to ever reach that much power but its also insanely good that it can reach that and survive !