TR Forums

Well this certainly seems rather interesting, what do the gerbils think.

Ryzen

I think a username like "credible" coupled to plagiarized drivel at wccftech is either trolling or has a deep sense of irony. In either event I've flagged your post.

There are times I like wccftech but they do often republish the same tables and charts and info and...

This isn't even info really, it's a recap of the event and there's hardly anything there.

Really Chuck, this is something I am 'known' for, please do use some common sense. 

I just happened across the article and thought folks here might like to look into it as I searched and saw nothing on TR.

I don't see a problem with this, but I can see discussions around the same event heading in the same direction as the other thread. If I see duplicate points being made, I may close this as a dupe. Right now, please focus on the numbers, info and speculation as a result of the event. Leave the personal jabs out of this thread.

/mod

There isn't anything new here because it's just a recap of the Ryzen live demo. There's Blender, Handbrake, that laughable Dota 2 demo...and nothing new to talk about, so I don't see how there can be new points.

Except that table at the end claiming a 4C/8T Ryzen variant for $149. If that comes to pass, then AMD missed their internal performance targets. Just no way.

Yeah, I skimmed through the page and noticed that also.  I figured it was just WTFTech wccftech adding a little extra "padding" to the story where they thought necessary.  Maybe there was some rumor reported which stated that I haven't really tried to keep up with it all, but if it's true it's not good news.

derFunkenstein wrote:

Except that table at the end claiming a 4C/8T Ryzen variant for $149. If that comes to pass, then AMD missed their internal performance targets. Just no way.

If the table is right, it is intended to be the "entry level" SKU of the range. Less cache, and (presumably) a low-ish base clock, and (probably) a locked multiplier. I could actually see that going for $149. It also gives AMD something better to do with dies that have multiple bad cores and/or defective cache than just throwing them in the trash.

And yes, I'm just pulling this out of my butt, because it's all still speculation at this point.

derFunkenstein wrote:

There isn't anything new here because it's just a recap of the Ryzen live demo. There's Blender, Handbrake, that laughable Dota 2 demo...and nothing new to talk about, so I don't see how there can be new points.

Except that table at the end claiming a 4C/8T Ryzen variant for $149. If that comes to pass, then AMD missed their internal performance targets. Just no way.

I like to see some heavy gaming benchmarks along with other applications that can use 8 cores 16 threads. What about Linux support?

derFunkenstein wrote:

There isn't anything new here because it's just a recap of the Ryzen live demo. There's Blender, Handbrake, that laughable Dota 2 demo...and nothing new to talk about, so I don't see how there can be new points.

Except that table at the end claiming a 4C/8T Ryzen variant for $149. If that comes to pass, then AMD missed their internal performance targets. Just no way.

It's the same table wccftech has been using in their articles for at least a week now (or so.)
I could see it being based on AMDs internal targets rather than actual final prices, but we'll have to see.

whm1974 wrote:

What about Linux support?

What about it?

Assuming they haven't screwed the pooch on x86-64 backward compatibility (which would be a fatal move), the CPU is supported.

Any new power management modes might take a little time to be fully supported.

Chipset support depends on whether they screwed up the USB and/or AHCI implementations, and when workarounds for any silicon bugs make it into the in-tree drivers.

System health monitoring and fan speed control is at the mercy of the motherboard vendors.

just brew it! wrote:

whm1974 wrote:

What about Linux support?

What about it?

Assuming they haven't screwed the pooch on x86-64 backward compatibility (which would be a fatal move), the CPU is supported.

Any new power management modes might take a little time to be fully supported.

Chipset support depends on whether they screwed up the USB and/or AHCI implementations, and when workarounds for any silicon bugs make it into the in-tree drivers.

System health monitoring and fan speed control is at the mercy of the motherboard vendors.

Well I really hope that the chipsets and motherboards will have decent Linux support.

chuckula wrote:

I think a username like "credible" coupled to plagiarized drivel at wccftech is either trolling or has a deep sense of irony. In either event I've flagged your post

I'm not sure what you're saying, are you claiming that the benchmarks he linked to aren't legit or that they are somehow "stolen" from someone or both?
It seems to me you're just looking for something to be angry about.

LostCat wrote:

derFunkenstein wrote:

There isn't anything new here because it's just a recap of the Ryzen live demo. There's Blender, Handbrake, that laughable Dota 2 demo...and nothing new to talk about, so I don't see how there can be new points.

Except that table at the end claiming a 4C/8T Ryzen variant for $149. If that comes to pass, then AMD missed their internal performance targets. Just no way.

It's the same table wccftech has been using in their articles for at least a week now (or so.)
I could see it being based on AMDs internal targets rather than actual final prices, but we'll have to see.

As a general rule I don't read the site. So in this case I don't mind showing my ignorance.

.cpchardware.com/cpc-hardware-n31-debarque-kiosque/

cdn.overclock.net/1/1a/1a431fe2_8a14207f-a115-4534-b6bd-c7801085ca42.jpeg

The same guy broke the K8 performance months before release.

Amiga500+ wrote:

.cpchardware.com/cpc-hardware-n31-debarque-kiosque/

cdn.overclock.net/1/1a/1a431fe2_8a14207f-a115-4534-b6bd-c7801085ca42.jpeg

The same guy broke the K8 performance months before release.

*IF* those benchmarks turn out to be accurate for the release silicon (and I think that's a big *if* given that at minimum these benches were done with pre-release chips so final release may have better clocks and optimization), it looks like it will perform very well at heavily threaded tasks like rendering and encoding, but it's getting beaten by the i5 6600 and even the Devil's Canyon i7 at games, all while using a fairly high amount of juice (only the 6700K and the FX chip used more). 
Not looking like a home run from those numbers -- maybe a double or a long single.

Well, much as wccftech is usually questionable, this article seems interesting.
Ryzen Review Leaked
It's claiming that a french magazine broke NDA, possibly by accident.
I do read french, and the shots do say what they said they do.
I haven't gotten a hold of the original article, so I can't speak to the nuances or anything else. Plus my technical french is kinda rusty to say the least.
If real, it's somewhat interesting, even if it is with non-final silicon.

EDIT: NMD I see this is the same thing referenced just earlier.

Looks like a very strong contender for threaded workloads, but single thread performance -- while a decent improvement -- is still kind of "meh" compared to Intel's current offerings. I suspect the "40% IPC gain" claim we've been hearing tossed around for a while may have been a "per core" IPC, not a "per thread" IPC -- i.e., one Zen core, running 2 threads with HT, is 40% faster than one (non-HT) Excavator core running a single thread.

If this info turns out to be accurate, Zen looks good enough to get me to bite (provided the platform is also decent), but not the "OMG, Intel is toast!" barn burner some of the more extreme fanboys have been claiming it would be.

Beating Intel was a stretch but that French source is supposedly legit as they broke NDA for previous AMD chips.

Within 3% of an i5 6600K with an engineering sample (and lower than production clocks) in gaming workloads is more than I ever dreamed. Performance/watt seems to be only slightly behind compared to Intel as well. I'd say if these leaks are true (and final clocks are 10% greater than the sample), this chip will have near Broadwell IPC (roughly 95% of) and power envelope while being somewhat lower in price. Best thing from AMD in a long, LONG time.

ultima_trev wrote:

Beating Intel was a stretch but that French source is supposedly legit as they broke NDA for previous AMD chips.

Within 3% of an i5 6600K with an engineering sample (and lower than production clocks) in gaming workloads is more than I ever dreamed. Performance/watt seems to be only slightly behind compared to Intel as well. I'd say if these leaks are true (and final clocks are 10% greater than the sample), this chip will have near Broadwell IPC (roughly 95% of) and power envelope while being somewhat lower in price. Best thing from AMD in a long, LONG time.

Isn't this an 8C/16T CPU, within 3 percent of a 4C/4T Intel offering?

I still think it's hard to draw real conclusions on this stuff. Once again, please let's just wait for real reviews.

As the source is known to be legit, speculation can't hurt.

As the i7 6700K and i7 4790K were ahead of the i7 6800K and i7 6900K in this test, it would seem while cores can make some difference it seems clocks still are more important.  The Zen engineering sample was 11% behind the 6900K in the gaming test however the Zen ES clock speeds were 6% lower, thus it would seem it is indeed within 5-6% IPC to Broadwell.

I used to like WCCF for my tech tabloid fix but not long ago a reader posted something about questioning the credibility of the site (and well, who takes them seriously anyway?). It wasn't particularly vulgar or very derogatory or anything but the way one of the site's writers hit back at the reader was deplorable as he totally bashed the reader with all manner of rude and derogatory remarks and counter arguments. Now I don't really consider WCCF to be very professional to begin with but after seeing that counter attack from the site all traces of being professional vaporized. I very seldom visit them anymore, and when I do it's just to skim over what gibberish they have.

I have a hard time believing AMD would supply samples to someone who had proven to disregard NDAs in the past, and even if that were the case there's still very little information about the system in use and the settings used for each test.  I get it.  I really do.  I'm thinking about a Ryzen upgrade also if things turn out well here, but the end result is that we still don't know much more than we did a month ago.

Slower than 6600k in gaming,........... one best hope, by some magical method, it's easily, easily overclockable to 4.5ghz.   Since Kaby Lake looks like it might become the first CPU to get 5ghz for more than just 1 or 2% of overclockers...

just brew it! wrote:

Looks like a very strong contender for threaded workloads, but single thread performance -- while a decent improvement -- is still kind of "meh" compared to Intel's current offerings. I suspect the "40% IPC gain" claim we've been hearing tossed around for a while may have been a "per core" IPC, not a "per thread" IPC -- i.e., one Zen core, running 2 threads with HT, is 40% faster than one (non-HT) Excavator core running a single thread.

If this info turns out to be accurate, Zen looks good enough to get me to bite (provided the platform is also decent), but not the "OMG, Intel is toast!" barn burner some of the more extreme fanboys have been claiming it would be.

The IPC certainly is there, but IPC is, by definition, per clock. The clocks aren't there. That it stands up to the 6600K whilst running at a clear frequency disadvantage shows that the IPC is better than Intel's. Single-threaded performance is not better.
The results are exactly what I expected and what I've been saying for a long time. Practically equivalent to the 6600K in games, much better than the 6700K in multi-threaded. Overall, worth a $300-350 sticker price to me. Hard sell at $500, although the comparison with the 6800K is very favorable for "pro" or "workstation" uses.
The advantage of this design is that it should be great in server chips, which have nice fat margins. For single-threaded, I think v2 is likely to be much better. Don't forget that FX-8150 (the original bulldozer) ran at 3.6GHz while FX-8350, released less than a year later, ran at 4.0GHz. Competition is heating.

DancinJack wrote:

ultima_trev wrote:

Beating Intel was a stretch but that French source is supposedly legit as they broke NDA for previous AMD chips.

Within 3% of an i5 6600K with an engineering sample (and lower than production clocks) in gaming workloads is more than I ever dreamed.  Performance/watt seems to be only slightly behind compared to Intel as well.  I'd say if these leaks are true (and final clocks are 10% greater than the sample), this chip will have near Broadwell IPC (roughly 95% of) and power envelope while being somewhat lower in price.  Best thing from AMD in a long, LONG time.

Isn't this an 8C/16T CPU, within 3 percent of a 4C/4T Intel offering?  

I don't want to nitpick, but it's within 3% if the only possible use for the processor is gaming. In multi-threaded workloads the Zen chip clearly delivers.
But wait, I can do better. The i3 6320 ($150) is within 5% of the 6950X ($1700) for gaming. See for example: http://www.anandtech.com/bench/product/1783?vs=1730
Saying that the new chip is similar to the 6600K is an oversimplification, one that is potentially misleading. We'll see what the final reviews bring...

DancinJack wrote:

Isn't this an 8C/16T CPU, within 3 percent of a 4C/4T Intel offering?

...in workloads that don't come anywhere close to using 16 threads. So the huge advantage in core/thread count is not (much of) a factor.

DancinJack wrote:

I still think it's hard to draw real conclusions on this stuff. Once again, please let's just wait for real reviews.

Yes, none of it is official yet. But it is starting to look like some of the rumors have legs.

ultima_trev wrote:

Beating Intel was a stretch but that French source is supposedly legit as they broke NDA for previous AMD chips.

Within 3% of an i5 6600K with an engineering sample (and lower than production clocks) in gaming workloads is more than I ever dreamed.  Performance/watt seems to be only slightly behind compared to Intel as well.  I'd say if these leaks are true (and final clocks are 10% greater than the sample), this chip will have near Broadwell IPC (roughly 95% of) and power envelope while being somewhat lower in price.  Best thing from AMD in a long, LONG time.

Frankly, who really cares about the gaming scores.
It's the purely CPU MT scores that are worrying which only puts Zen in the SB IPC ballpark. 

strangerguy wrote:

Frankly, who really cares about the gaming scores.

Gamers? (Who also happen to represent a significant percentage of this site's readeship?)

I'd also be willing to bet that as more and more "everyday" computing has gone Cloud/mobile, gamers represent a larger share of a (shrinking) desktop PC market.

strangerguy wrote:

It's the purely CPU MT scores that are worrying which only puts Zen in the SB IPC ballpark.

A home run would've been nice. They hit a double instead. At least they didn't strike out. </baseball-analogy>

strangerguy wrote:

It's the purely CPU MT scores that are worrying which only puts Zen in the SB IPC ballpark. 

It definitely looks closer to Haswell or Ivy Bridge than Sandy Bridge.  Keep in mind the Intel chip is clocked 6% higher; The i7 6900K score 193.4 overall in the multi threaded test versus 168.7 for the Zen ES.  Since it's not unreasonable for Zen's final production clocks be 6% higher, there's no reason for it to score less than 175-180 with production clocks.  Let's say it scores 177.4...

177.4 / 193.4 = 0.9172~

So 91.7 percent of Broadwell-E performance per clock in multi-threaded workloads.  Also the rumored price for the top Summit Ridge Ryzen is about $500, so if true AMD will be quite close to doubling the price/performance of the i7 6900K.