Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

AMD Zen and Ryzen 7 Review: A Deep Dive

anandtech.com

Post reply on HN
Print view View original

81–90 of 359 posts

Re: AMD Zen and Ryzen 7 Review: A Deep Dive

#81

Post by phamilton »

Tom1971 wrote: post #57
AnthonyMouse wrote: post #51

Earlier quoted context omitted.

> But it's still losing in every web and office single threaded benchmark. By ~15% compared to >160% for Bulldozer. And this is day zero before any applications have been optimized for the new microarchitecture, compared with Core which everything has been targeting for more than a decade. They did good.

I don't think spending time and effort to optimize code for Ryzen will be time well spent. Would you do it if you had the choice between that or adding new money making features/fixing bugs?

This is a compiler optimization, not an application optimization. I'd certainly add -march=zen as a compiler flag for my application.

As for the priorities of the compiler devs, I think it will get some love.

Re: AMD Zen and Ryzen 7 Review: A Deep Dive

#82

Post by robotjosh »

AnthonyMouse wrote: post #51
yread wrote: post #4

Earlier quoted context omitted.

But it's still losing in every web and office single threaded benchmark. EDIT: compared to 7700k that costs about the same

> But it's still losing in every web and office single threaded benchmark. By ~15% compared to >160% for Bulldozer. And this is day zero before any applications have been optimized for the new microarchitecture, compared with Core which everything has been targeting for more than a decade. They did good.

Good reason to wait for the next chip cycle. AMD will have smoothed out any rough edges and Intel will show us what they've really got.

Re: AMD Zen and Ryzen 7 Review: A Deep Dive

#84

Post by dijit »

bdz wrote: post #72

Looking at those gaming benchmarks. Turns out the single core performance is still more important than having more cores. https://www.purepc.pl/procesory/premiera_i_test_procesora_am...

main thread has to go somewhere. :\

Maybe knowing consumers will probably have more than 2 cores will make games optimise better for parellelism, it's not like the engine programmers _cant_ do it, they do it all the time with GPUs.

Re: AMD Zen and Ryzen 7 Review: A Deep Dive

#85

Post by JamesMcMinn »

Tom1971 wrote: post #57
AnthonyMouse wrote: post #51

Earlier quoted context omitted.

> But it's still losing in every web and office single threaded benchmark. By ~15% compared to >160% for Bulldozer. And this is day zero before any applications have been optimized for the new microarchitecture, compared with Core which everything has been targeting for more than a decade. They did good.

I don't think spending time and effort to optimize code for Ryzen will be time well spent. Would you do it if you had the choice between that or adding new money making features/fixing bugs?

Give compilers some time and the optimizations will be "free" to developers.

Besides, if your application is one where performance is a serious issue, then you are going to spend some time optimizing rather than adding features. For some applications, performance is a feature.

Re: AMD Zen and Ryzen 7 Review: A Deep Dive

#86

Post by lorenzhs »

Tom1971 wrote: post #57
AnthonyMouse wrote: post #51

Earlier quoted context omitted.

> But it's still losing in every web and office single threaded benchmark. By ~15% compared to >160% for Bulldozer. And this is day zero before any applications have been optimized for the new microarchitecture, compared with Core which everything has been targeting for more than a decade. They did good.

I don't think spending time and effort to optimize code for Ryzen will be time well spent. Would you do it if you had the choice between that or adding new money making features/fixing bugs?

It is for compiler authors, or libc etc. There's very little most application authors would need to do.

Re: AMD Zen and Ryzen 7 Review: A Deep Dive

#87

Post by phamilton »

Asooka wrote: post #59
snovv_crash wrote: post #28

Earlier quoted context omitted.

It's also up to 30% faster. That's how variance works.

Well yeah but will you buy an alternate CPU that might be faster or might be slower in some games, or will you buy standard Intel, for which every game was definitely tested and will run with standard performance?

Frame of reference fallacy going on here. One could just as easily say the Intel is 30% faster or slower.

Re: AMD Zen and Ryzen 7 Review: A Deep Dive

#88

Post by yborg »

qaq wrote: post #62
anthonybsd wrote: post #40

Earlier quoted context omitted.

Consumer applications? Doubt it. Not a whole lot of consumer-oriented applications parallelize that well to be honest.

look at how many threads your Chrome is running

All but one set (for the active tab) of which are probably idle or effectively so most of the time.

Re: AMD Zen and Ryzen 7 Review: A Deep Dive

#89

Post by slantyyz »

jdietrich wrote: post #75
yread wrote: post #4

Earlier quoted context omitted.

But it's still losing in every web and office single threaded benchmark. EDIT: compared to 7700k that costs about the same

The R7 1700 is within shouting distance of the 6900k in most serious tasks - rendering, encoding, compilation etc. A $330 65w CPU is a credible alternative to an $1100 140w CPU. That is absolutely remarkable. The cherry on the cake is that the cheapest AM4 motherboards are half the price of the cheapest 2011-v3 boards. For the first time in a decade, AMD are seriously competing at the high end. That's great news for…

I'm hoping things get more interesting when the Ryzen Mobile chips come out later in the year.

Re: AMD Zen and Ryzen 7 Review: A Deep Dive

#90

Post by Symmetry »

tutanchamun wrote: post #47
arcanus wrote: post #27

Earlier quoted context omitted.

Yep. And nvidia and amd are hurting Intel in terms of servers which are moving in many cases towards GPU for acceleration. Very interesting times in Silicon.

Yeah, and AMD is the only company who can offer a high performance GPU together with a high performance x86 CPU. And APU's with HBM of course. Looking forward to what they can achieve now that they seem to be back (although most of the financial success probably lies with the server CPU Naples).

One can hope that in the long run we'll see more things with a SPRI-V back end and more libraries in that space to put AMD and NVidia GPUs on more equal footing.
Post reply on HN