Pieter3dnow

Supreme [H]ardness
Joined
Jul 29, 2009
Messages
6,784
Here Jim goes where he has not gone before with Zen 2 trying to explain what we have seen so far and reflection on how AMD uses benchmarks to profoundly show their best case scenario for performance.

 
I didnt watch but based on what you wrote, is the video implying that Intel doesnt do the same?
 
I didnt watch but based on what you wrote, is the video implying that Intel doesnt do the same?
No it is completely based on what AMD does. It does have reflection on what AMD does compared to others but in mention only not indepth....
 
I didnt watch but based on what you wrote, is the video implying that Intel doesnt do the same?

I watched it, it's just basically a review of what we can likely expect from Zen two given what was shown of Zen leading to release, and how engineering samples have compared to release versions and how amd has used demonstration benchmarks in the past to show their products in the absolute best way possible. Nothing that really merits it being mentioned here in its own thread.
 
Actually it is very important that people understand this.
Coming here for a good decade I'm not even sure that most people do get this ...
 
  • Like
Reactions: otg
like this
Well car companies do this. Clinton, Trump, candy bars, soap, cigarettes, beer, sex products, pharma.

I'm not sure what's different that AMD is doing.
 
But it is more technical then anything else. It shows a lot of things where you could think that Polaris is the best gpu ever and see what it really does makes it not that attractive.

I found the cinebench stuff a little bit confusing the issue since that has been abused by Intel to stomp on Bulldozer. Remember that benchmark inside CPUID and how that got fixed when Zen launched because it used a gimmick to calculate by pausing it slightly. It shows how weird and complicated things get.
 
I'm not sure i understand. Is the premise that AMD should show their worst case scenario? Does this hold true for all other companies?
 
I'm not sure i understand. Is the premise that AMD should show their worst case scenario? Does this hold true for all other companies?

Are you talking about the video, or about Pieter3dnow? If you mean the video, it was just about assessing what AMD has shown compared to what they usually show. They usually show the most favorable thing they can. In polaris ten they showed a very purposely controlled scenario to make it seem like their architecture was more efficient than maxwell. In Zen they showed the right scenario to make it seem like it was as strong as broadwell-E, which it arguably is. In Zen2 they showed how energy efficient Zen2 is compared to Intel's latest while doing certain multithreaded workloads. There's no reason to believe the new chips aren't that efficient at that workload, it just doesn't say anything else about any other performance characteristics.

We already know the Zen architecture is good enough to compete with Intel at most things. What we haven't seen so far is anything to indicate Zen2 will be the "best" CPU for posting benchmark scores on counterstrike at 720p with soup textures.
 
We already know the Zen architecture is good enough to compete with Intel at most things. What we haven't seen so far is anything to indicate Zen2 will be the "best" CPU for posting benchmark scores on counterstrike at 720p with soup textures.

That is a metric which is based around moving things in and out of cache memory really fast. It can be as meaning less as if you want to test compression on a large file only containing the letter A repeatedly.
 
My take after watching this was simply Jim playing "Devil's Advocate" and attempting to lay out a scenario where his sources/leakers are completely wrong and thus "Zen 2 will suck" but in the wrap-up, Jim states:

"...I was talking to an AMD PR guy recently, and I told him I was planning this video, and he said to me "Jim, I think you're going to have a challenge to say Zen 2 will be bad." Well, I tried, and I just can't suspend my belief much further than I've been doing all video. I speculated on how Zen 2 may not succeed, but with the info we have, I feel I'm making multiple stretches to justify that [Zen 2 wil be bad/suck/Zen 2 may not succeed] possibility...."

"...What I would say is that I think there's a decent chance that AMD does choose to launch the 12-core parts first of all, with 16-core parts coming later. It could even well be that they launch this 8-core part first of all, but I believe that would be a mistake. But for me this [only launching 8-cores initially, with 12 and 16-core coming later] is the most likely "bad news" regarding Zen 2 and it wouldn't be done for technical reasons but for marketing and segmentation reasons. We'd get the [more than 8-core] chips eventually, but they'd be rolled out ever-faster parts over a few months...."
 
  • Like
Reactions: otg
like this
My take after watching this was simply Jim playing "Devil's Advocate" and attempting to lay out a scenario where his sources/leakers are completely wrong and thus "Zen 2 will suck" but in the wrap-up, Jim states:

"...I was talking to an AMD PR guy recently, and I told him I was planning this video, and he said to me "Jim, I think you're going to have a challenge to say Zen 2 will be bad." Well, I tried, and I just can't suspend my belief much further than I've been doing all video. I speculated on how Zen 2 may not succeed, but with the info we have, I feel I'm making multiple stretches to justify that [Zen 2 wil be bad/suck/Zen 2 may not succeed] possibility...."

"...What I would say is that I think there's a decent chance that AMD does choose to launch the 12-core parts first of all, with 16-core parts coming later. It could even well be that they launch this 8-core part first of all, but I believe that would be a mistake. But for me this [only launching 8-cores initially, with 12 and 16-core coming later] is the most likely "bad news" regarding Zen 2 and it wouldn't be done for technical reasons but for marketing and segmentation reasons. We'd get the [more than 8-core] chips eventually, but they'd be rolled out ever-faster parts over a few months...."

Yeah, I guess most people didn't watch it or watch it through.
 
  • Like
Reactions: otg
like this
Are you talking about the video, or about Pieter3dnow? If you mean the video, it was just about assessing what AMD has shown compared to what they usually show. They usually show the most favorable thing they can. In polaris ten they showed a very purposely controlled scenario to make it seem like their architecture was more efficient than maxwell. In Zen they showed the right scenario to make it seem like it was as strong as broadwell-E, which it arguably is. In Zen2 they showed how energy efficient Zen2 is compared to Intel's latest while doing certain multithreaded workloads. There's no reason to believe the new chips aren't that efficient at that workload, it just doesn't say anything else about any other performance characteristics.

We already know the Zen architecture is good enough to compete with Intel at most things. What we haven't seen so far is anything to indicate Zen2 will be the "best" CPU for posting benchmark scores on counterstrike at 720p with soup textures.

Well ok, but i would have thought all that was rather obvious. So about the same as nvidia saying X architecture is 10x faster than Y architecture. I hope this isn't cementing the notion that people need to be spoon fed information in one direction or another to analyze these kinds of things. It says more about society in this day and age as it does anything else really.
 
Well ok, but i would have thought all that was rather obvious. So about the same as nvidia saying X architecture is 10x faster than Y architecture. I hope this isn't cementing the notion that people need to be spoon fed information in one direction or another to analyze these kinds of things. It says more about society in this day and age as it does anything else really.

Have you hung around this, or any other, tech forum, around the time of a major product launch?
Even alleged 'enthusiasts' seem to need a healthy dose of the "rather obvious".
 
Back
Top