• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Ryzen "2" ?

Ah you beat me too it ^^^^^ :p

Mercutio said:
Another game benchmark:

fpzR1b8.png

+ reddit thread:
https://www.reddit.com/r/Amd/comments/87ok0m/saw_this_photo_of_a_comparison_between_2700x_and/
Click to expand...

That looks a lot more like it.....

CPU Frame Rate:


1800X: 96.8
2700X: 123.3 (27%)

1800X: 105.8
2700X: 138.1 (+30%)

1800X: 101.7
2700X: 127.9 (+26%)

1800X: 85.4
2700X: 107.8 (+26%)

So pretty consistently about +25%, that's a lot more than i was expecting but because it actually pushes the them hard this game does pretty heavily favors AMD CPU's, i'm not expecting near 30% for other games but this contradicts CanarPC completely.
 
This will be because boost has better granularity on new chips, with many threads clockspeed will be higher than the old chips, still doesn't mean single thread is any good.

Still even Intel are pushing multithread game design now as they can't get their clocks much higher and extra cores will be order of the day, so we may finally see a shift away from poorly threaded game design. A win for everybody.
 
humbug said:
Its where the Application deliberately throws a lot of Draw Calls at the CPU, enough to insure a complete bottleneck of the GPU so as to test the performance of the CPU. that is where AMD shine, under very high load.
Click to expand...

To put that into perspective, ATOS is to AMD what Cinebench used to be to Intel, before Ryzen came along....

So don't be expecting the same level of uplifts for all games but certainly it does contradict CanardPC because you don't get near +30% unless your CPU is actually faster, even if the CPU loves the game anyway.

Iglttr3.png
 
Its not about the CPU cores, you may have noticed the 8700K is not a lot faster than the 7700K, the 1800X not a lot faster than the 1800X.

This is about high load Floating Point performance, Ryzen, all Ryzen 1### CPU's have about 20MB of Cache, the 8700K about 13MB, the 7700K about 11MB, when a CPU is doing a huge amount of little bits of work the more of that work it can run A-Synchronously in Cache the faster it is, Ryzen CPU's have a lot more Cache than Intel CPU's, probably for the same reason Intel use Cheap paste to bond their Heat Spreaders to the CPU DIE Cap, Cache is also expensive, it takes up a lot of space on the CPU.

Intel only really care about light serial workloads because that's what most benchmarks are.

Having said that i think the 6900 does have more cache than the 8700K.

Edit, Threadripper has about 40MB of Cache.
 
4K8KW10 said:
10-core / 20-thread CPU must be faster http://www.tomshardware.co.uk/intel-core-i7-7820x-skylake-x,review-33961-4.html
Click to expand...

Typical Toms Hardware, everything about this seems right, the 7900X topping the board, yeah... apart from one thing, the 7700K, it look like they took that from a previous benchmark that they pushed to the top of the charts surrounded by Intel advert banners and forgot they did that, only this is a Skylake-X review so the 7900X had to be faster and yet the 4.6Ghz 7820X is near at the bottom with the 4Ghz 1600X. is that the true performance of Skylake-X, how is the stock 7700K so much faster?

Every time is see something from them it stinks, they do anything for advertising revenue and Intel pay big.

lNoto9Y.jpg.png
 
sandys said:
This will be because boost has better granularity on new chips, with many threads clockspeed will be higher than the old chips, still doesn't mean single thread is any good.

Still even Intel are pushing multithread game design now as they can't get their clocks much higher and extra cores will be order of the day, so we may finally see a shift away from poorly threaded game design. A win for everybody.
Click to expand...
The topic of intel pushing more cores starts around 10:58. A bit later it is mentioned that at the 10nm process, intel can't get the clock speed to match their current 14nm, which I thought was odd.
https://www.youtube.com/watch?v=WYvOhJsO71I
 
Trek9.9sl said:
The topic of intel pushing more cores starts around 10:58. A bit later it is mentioned that at the 10nm process, intel can't get the clock speed to match their current 14nm, which I thought was odd.
https://www.youtube.com/watch?v=WYvOhJsO71I
Click to expand...
Pretty sure it's been known for a while that Intel were having issues with clock speeds on their 10 nm node. They've definitely said before that they expect Cannonlake to be slower than Kaby Lake (albeit more power efficient, one would hope). They could counteract this by improving the architecture to increase IPC but they haven't done this in years so maybe they are struggling to get anything out of what they have currently. Maybe they'll go for a fresh approach like AMD did with Zen but if they haven't started yet it'll take a long time.
 
DragonQ said:
Pretty sure it's been known for a while that Intel were having issues with clock speeds on their 10 nm node. They've definitely said before that they expect Cannonlake to be slower than Kaby Lake (albeit more power efficient, one would hope). They could counteract this by improving the architecture to increase IPC but they haven't done this in years so maybe they are struggling to get anything out of what they have currently. Maybe they'll go for a fresh approach like AMD did with Zen but if they haven't started yet it'll take a long time.
Click to expand...

Intel were advertising for a "next gen CPU architect" type person around the time Zen came out. Probably safe to assume they got a relevant person relatively quickly... they'll still only be about 10 months into developing a new architecture if one was coming. We won't hear about it for a good few years yet.
 
DragonQ said:
Pretty sure it's been known for a while that Intel were having issues with clock speeds on their 10 nm node. They've definitely said before that they expect Cannonlake to be slower than Kaby Lake (albeit more power efficient, one would hope). They could counteract this by improving the architecture to increase IPC but they haven't done this in years so maybe they are struggling to get anything out of what they have currently. Maybe they'll go for a fresh approach like AMD did with Zen but if they haven't started yet it'll take a long time.
Click to expand...

Bit of a rock and a hard place for Intel really, they have been able to sit pretty for years knowing they had a clockspeed and IPC lead over AMD, when AMD released stuff that could clock to 5ghz it was miserable in actual performance, so Intel was still sitting pretty, as their stuff still was king of the heap.

Now though it seems like they have stagnated too long, as AMD are catching them up, AMD continued the trend of more cores, but managed to get a lot closer with IPC, now it seems Intel is struggling with their forthcoming shrinks, AMD are right on their tails.

Right now, if Intel could lob 2 more cores on an 8700k , keep the same thermals and lead in single threaded perf over Ryzen it would clean up, but i cant see that happening, adding more cores as we all know is going to spike the power and heat up, to counteract this they will need to reduce something to keep it tamed.

I fully expect Ryzen 2 (Not Ryzen+) to really put weight on Intel, especially if it crops up next year, Intel's only real route of progression right now is an 8700k with 2 more cores, keeping the same IPC etc and dealing with Thermals, people will lap them up, i cant see them increasing their IPC lead any further right now especially if they are moving towards higher core count cpu's.

What will probably happen is they will end up ironically doing what AMD done, and bunging 2 packages together in a single chip etc, unfortunately for them, AMD have the lead on that setup right now.

Its going to be interesting times over the next 2 -3 years in the CPU arena.
 
SiDeards73 said:
...adding more cores as we all know is going to spike the power and heat up, to counteract this they will need to reduce something to keep it tamed.
Click to expand...

You've made me wonder if the 8700K might actually be the fastest single-thread processor we'll see for the next few years... even soldering the spreader probably won't let them handle 8 Coffee cores on any normal sort of cooling...
 
DragonQ said:
Pretty sure it's been known for a while that Intel were having issues with clock speeds on their 10 nm node. They've definitely said before that they expect Cannonlake to be slower than Kaby Lake (albeit more power efficient, one would hope). They could counteract this by improving the architecture to increase IPC but they haven't done this in years so maybe they are struggling to get anything out of what they have currently. Maybe they'll go for a fresh approach like AMD did with Zen but if they haven't started yet it'll take a long time.
Click to expand...

I dunno what Intel are playing at with 10nm not that it is trivial working at those levels and the lithography extremely challenging but they've twice choked when it has come to executing on what others like Samsung have managed and seem to be really struggling with EUV which is crucial beyond this point. Intel had no intention of delaying 10nm this long or still using processes based on 20nm planar for their top end lines at this point - their 10nm papers show it ready for products in the later half of 2017.
 
You must log in or register to reply here.
Back
Top Bottom