"12nm" is just 14nm+. GlobalFoundries called it "12nm" because it sounds better.
I also said months ago that if GlobalFoundries promise 10% performance uplift, expect half of that.
It looks like there is nice performance improvement even accounting for the clock speed.
That's actually incredible for a refresh.It looks like there is nice performance improvement even accounting for the clock speed.
Yes, that's probably the case. I get almost exactly the same Cinebench score with 2133 MHz and 3466Mhz on my 1700X.At the same clock, Cinebench only went up 3% while games sent up 9-14%
So, it looks like Cinebench is not very sensitive to memory latency, while games are much more so.
Yeah, people in reddit are claiming the Cinebench score should be 1765 - 1782. IMO this is blown out of proportion and is well within the margin of error (1-2%). And it's not like the 173x score would be an outlier, techpowerup got a similar score and I'm sure there are others. Be that as it might, those single-digit percentage differences won't convert into any meaningful difference in games anyway.I heard some people say that the 1700 is underperforming in that review, supposedly the multi core CB score is low in particular, but I'm not sure if it has truth to it. Hopefully, it doesn't, I guess Ryzen owners here would know.
It's as we expected, games are very latency bound and we should see a big performance increase in them. As raghu said, the all core OC is somewhat disappointing because it seems that AMD went for a direct port to 12nm. Still great for a refresh like this. Now I'm waiting on that eCLK and XFR 2.0 overclocking to see if we will have any more nice surprises.
This "quick-n-dirty refresh" has more per-clock gains over Zen than Intel has had in three years.quick-n-dirty refresh
KitGuru got the 2700X score at 4.1GHz on their 1800X.Yeah, people in reddit are claiming the Cinebench score should be 1765 - 1782. IMO this is blown out of proportion and is well within the margin of error (1-2%). And it's not like the 173x score would be an outlier, techpowerup got a similar score and I'm sure there are others. Be that as it might, those single-digit percentage differences won't convert into any meaningful difference in games anyway.
The OC is a tad disappointing for sure, but I wouldn't blame AMD for not wasting tons of resources on a quick-n-dirty refresh when they should be working on Zen 2. I bet the new process (with quad-patterning and crazy amount of masks, etc) requires tons of work, to design to. Do you really want a zen+ with 100-200 extra Mhz at the expense of Zen2? Say, the very same decision causes a 3 month delay on Zen2, all while 8-core 10nm intel XXX-Lakes are out (with slight IPC improvements). I bet the very same people would be bitching a whole lot more then.
Right, I thought so. And I agree, the potential frequency uplift was probably not worth a delayed launch and a lot more work. I think this is more than enough to bridge the gap between Zen and Zen 2. Honestly tempted to just treat myself to a 2700X right now, but I think I'll wait for Zen 2.Yeah, people in reddit are claiming the Cinebench score should be 1765 - 1782. IMO this is blown out of proportion and is well within the margin of error (1-2%). And it's not like the 173x score would be an outlier, techpowerup got a similar score and I'm sure there are others. Be that as it might, those single-digit percentage differences won't convert into any meaningful difference in games anyway.
The OC is a tad disappointing for sure, but I wouldn't blame AMD for not wasting tons of resources on a quick-n-dirty refresh when they should be working on Zen 2. I bet the new process (with quad-patterning and crazy amount of masks, etc) requires tons of work, to design to. Do you really want a zen+ with 100-200 extra Mhz at the expense of Zen2? Say, the very same decision causes a 3 month delay on Zen2, all while 8-core 10nm intel XXX-Lakes are out (with slight IPC improvements). I bet the very same people would be bitching a whole lot more then.
And certain people on a different site were telling me that this is just first gen Ryzen with an overclock after the leaks started coming out, in spite of all of the other improvements which were apparent...This "quick-n-dirty refresh" has more per-clock gains over Zen than Intel has had in three years.
Do you really want a zen+ with 100-200 extra Mhz at the expense of Zen2? Say, the very same decision causes a 3 month delay on Zen2, all while 8-core 10nm intel XXX-Lakes are out (with slight IPC improvements). I bet the very same people would be bitching a whole lot more then.
Well now we know what "refresh" means for AMD CPU team.And certain people on a different site were telling me that this is just first gen Ryzen with an overclock after the leaks started coming out, in spite of all of the other improvements which were apparent...
Exactly.
AMD have made the right move. Ryzen 2XXX looks like a solid step, fixing most of the major issues in Ryzen 1XXX.
Always easier to make gains when your starting point is lower.This "quick-n-dirty refresh" has more per-clock gains over Zen than Intel has had in three years.
They were kinda sorta near SKL anyway and now it's basically parity.Always easier to make gains when your starting point is lower.
No. Zen was close to Haswell in IPC and slower in gaming and memory latency sensitive applications. Zen+ can be considered on par with Broadwell.They were kinda sorta near SKL anyway and now it's basically parity.
Besides, it's been 3 years and Intel is still shipping SKL, refusing to backport CNL and Gen10 IP to 14+/++.
Pathetic.
Well let's just wait until someone locks 2600X and 8700K at the same clock and runs a bunch of benches.No. Zen was close to Haswell in IPC and slower in gaming and memory latency sensitive applications. Zen+ can be considered on par with Broadwell.
Let's wait for more benchmarks. In any case, the gains for intel have been from clock speed increases, not ipc, especially in mobile and at stock clocks on desktop.This "quick-n-dirty refresh" has more per-clock gains over Zen than Intel has had in three years.
But as expected, intel is excoriated for a 5 to 10 percent (more in mobile, actually) gain, while it is called a great improvement for AMD.
We're talking gains per clock in a single refresh.But as expected, intel is excoriated for a 5 to 10 percent (more in mobile, actually) gain
Now now, let's not go to ridiculous territory with the Intel Bashing. What about Ivy Bridge?We're talking gains per clock in a single refresh.
That's not something that Intel has done, ever.
I'm all for rooting for AMD. But constantly bashing and ridiculing Intel makes me think along the same lines as AdoredTV.Anand said:On the CPU core side that means you can expect clock-for-clock performance improvements in the 4 - 6% range.