|
AMD Zen details emerge-Dev opinions?
#357819 - 08/18/16 04:02 PM
|
|
|
|
|
Re: AMD Zen details emerge-Dev opinions?
[Re: DiodeDude]
#357879 - 08/20/16 12:42 PM
|
|
|
Not a Dev, but until we know just how high Zen can overclock/boost to, Intel should still be the default choice for Mame.
That demo had the chip running at only 3Ghz to illustrate that clock for clock it can rival Broadwell E which is all very good but if Zen can only reach 3-3.5Ghz, whilst the vast majority of Intel chips can easily go past 4.2Ghz, Intel will still be much faster at the end of the day.
|
|
|
|
Re: AMD Zen details emerge-Dev opinions?
[Re: DiodeDude]
#357881 - 08/20/16 04:51 PM
|
|
|
> What do you think?
I think you must be pretty non-technical indeed if you think that that article has enough detail to go on for anyone to form any sort of opinion. It's a bunch of marketing gibberish and zero real-world performance figures. Benchmark numbers or fuck off.
|
|
|
|
Re: AMD Zen details emerge-Dev opinions?
[Re: MooglyGuy]
#357884 - 08/20/16 05:40 PM
|
|
|
> > What do you think? > > I think you must be pretty non-technical indeed if you think that that article has > enough detail to go on for anyone to form any sort of opinion. It's a bunch of > marketing gibberish and zero real-world performance figures. Benchmark numbers or > fuck off.
I build,troubleshoot and repair computers. Yes I know that's nowhere near your level, so apologies for falling short of grace.
There are some supposed "leaked" benchmarks of an engineering sample floating around, but its obviously not emulation oriented.
Another fluff article for informational purposes:
http://arstechnica.com/gadgets/2016/08/amd-zen-performance-details-release-date/
|
|
|
R. Belmont |
Cuckoo for IGAvania
|
|
|
Reged: 09/21/03
|
Posts: 9717
|
Loc: ECV-197 The Orville
|
|
Send PM
|
|
|
Re: AMD Zen details emerge-Dev opinions?
[Re: DiodeDude]
#357921 - 08/22/16 06:55 PM
|
|
|
*If* the performance claims hold up (and AMD has a poor recent history on this front; they claimed the "Hammer" chips would bury Sandy Bridge, which they didn't), they'll actually be competitive with Intel for the first time in around 10 years. But I'll wait until review samples are available.
And Ziggy makes an excellent point that even if they match Skylake cycle-per-cycle they still lose if they're limited to 3 GHz.
|
|
|
|
Re: AMD Zen details emerge-Dev opinions?
[Re: R. Belmont]
#357927 - 08/22/16 11:31 PM
|
|
|
> *If* the performance claims hold up (and AMD has a poor recent history on this front; > they claimed the "Hammer" chips would bury Sandy Bridge, which they didn't), they'll > actually be competitive with Intel for the first time in around 10 years. But I'll > wait until review samples are available. > > And Ziggy makes an excellent point that even if they match Skylake cycle-per-cycle > they still lose if they're limited to 3 GHz.
Exactly. Suppose they do match Broadwell clock-for-clock. Then they are ~3% slower than the current generation skylake processors per clock. And then probably a few more percent behind whatever intel actually has out when Zen comes out.
That's still much better than they've done in years. But to actually be competitive they'll need to be competitive in clock speed too.
I mean, Intel could release skylake processors at 4.5GHz tomorrow if they wanted to, right? They just haven't had a need to push things like that.
Of course, even if AMD could only put out a 3.2-3.5GHZ Zen processor next year, it wouldn't be a bad CPU to have. It just wouldn't compete with the higher end offerings from intel. But you wouldn't have to give dire warnings to people about the need to avoid AMD for certain kinds of computing tasks (like emulation) anymore.
Unfortunately everything depends on how much AMD is lying about how good Zen is. Because I wouldn't be too surprised if it turned out that: (1) AMD used a benchmark that is unusually high performing on Zen (2) AMD used an old buggy version of the benchmark that has a performance issue for the intel cpu in the comparison (3) AMD actually was running some custom compute code on a gpu integrated in the cpu to get high benchmark scores, but of course that doesn't apply to any real applications you'd run. (4) AMD needed liquid nitrogen to get 3GHz and the real thing will be released as a "Zen 4500+" that runs at 2.2GHz.
Now some of those are jokes, but AMD doesn't have a good track record on this stuff lately.
Of course, Zen really does look like a much improved architecture so let's all hope for the best!
|
|
|
R. Belmont |
Cuckoo for IGAvania
|
|
|
Reged: 09/21/03
|
Posts: 9717
|
Loc: ECV-197 The Orville
|
|
Send PM
|
|
|
Re: AMD Zen details emerge-Dev opinions?
[Re: SoltanGris42]
#357969 - 08/24/16 05:49 PM
|
|
|
> Of course, Zen really does look like a much improved architecture so let's all hope > for the best!
Absolutely. I *want* them to do better. Intel needs some competitive pressure on them again.
|
|
|