Soldato
- Joined
- 22 Nov 2018
- Posts
- 2,840
Well according to Humbug's bar chart above, Intel has a 3.57% IPC lead over Zen+ which itself has a 2.44% gain over the original Zen. Zen 2 has to have a larger IPC gain than 5% though surely?
-
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.
AMD Zen 2 (Ryzen 3000) - *** NO COMPETITOR HINTING ***
- Thread starter Jamin280672
- Start date
Well according to Humbug's bar chart above, Intel has a 3.57% IPC lead over Zen+ which itself has a 2.44% gain over the original Zen. Zen 2 has to have a larger IPC gain than 5% though surely?
Caporegime
- Joined
- 18 Sep 2009
- Posts
- 30,292
- Location
- Dormanstown.
The multi-threaded performance of the Ryzen 2600X at 4GHZ is higher than that of the i7 8700K at 4GHZ. Core for core the Intel has higher IPC per single core, but when it's 1 Core 2 thread it's a different result, SMT is superior to HT in performance. This would be true with the Cinebench result that AMD showed.
The frame interval graph for the 9900k doesn't look to good imho.. sure it has higher average but look at the spikes. Surely it cant be that bad normally?
I just watched some of it, sometimes the 9900K has better frame times, other times its the 2700X... i think what that says is they are different CPU's.
Shadow Of The Tomb Raider does start out a bit stuttery to begin with on the 2700X, just the first few seconds before settling down, loading in stutter?
Associate
- Joined
- 2 Jan 2019
- Posts
- 617
I'd be a little wary of direct comparisons with a 9900K @4.7GHz. Whilst that is it's all core turbo, and most Mobos will have had their turbo durations set to infinite, there are two potential contaminants as far as judging IPC goes; firstly, turbo isn't supposed to be sustained indefinitely and the 9900K typically clocks down to 4.2GHz when performing to spec, whilst the XFR2 on the 2700X pretty much does guarantee that it'll run at 4.025GHz all core indefinitely, and secondly, games will certainly see a benefit from the single core boosts, though in fairness they do both typically boost a similar amount above their all core turbo clocks.
Before taking any comparison as being representative, I'd want to know exactly what clocks are being hit at all times.
Before taking any comparison as being representative, I'd want to know exactly what clocks are being hit at all times.
The Clocks are displayed in the MSI OSD through out.
Associate
- Joined
- 2 Jan 2019
- Posts
- 617
OK thanks.
Just watched the video.
Barring 3 outliers (2 in favour of Intel, 1 in favour of AMD), the pretty consistent fps difference was +10-12% in favour of the 9900K.
On a clock for clock basis that would have the 2700X already ahead of the 9900K, though fps gains tend not to be linear with clocks, so I would have the 2700X on a par with the 9900K already.
Any gains in ST and clocks, for the 3000 SKUs, would put them comfortably ahead IMO; as in, above margin of error ahead.
Just watched the video.
Barring 3 outliers (2 in favour of Intel, 1 in favour of AMD), the pretty consistent fps difference was +10-12% in favour of the 9900K.
On a clock for clock basis that would have the 2700X already ahead of the 9900K, though fps gains tend not to be linear with clocks, so I would have the 2700X on a par with the 9900K already.
Any gains in ST and clocks, for the 3000 SKUs, would put them comfortably ahead IMO; as in, above margin of error ahead.
I have to say I have a few concerns/reservations regarding the 3000 series.
They say they have changed the architecture from Zen/Zen+. This concerns me more so on gaming performance. If it's a significant departure from Zen 1 design, then it will take developers time to get used to coding for the new architecture.
Does it still resemble a "mesh" architecture? In which case it won't be as efficient as Intels ring architecture for gaming.
You have to remember, Intel has used a ring architecture for a very long time, and they haven't really changed their architecture significantly since sandybridge. This meant, with every new Intel CPU release, developers haven't really needed to alter code much for Intel. It's mainly been about clock speed.
Hopefully I'm wrong though!
They say they have changed the architecture from Zen/Zen+. This concerns me more so on gaming performance. If it's a significant departure from Zen 1 design, then it will take developers time to get used to coding for the new architecture.
Does it still resemble a "mesh" architecture? In which case it won't be as efficient as Intels ring architecture for gaming.
You have to remember, Intel has used a ring architecture for a very long time, and they haven't really changed their architecture significantly since sandybridge. This meant, with every new Intel CPU release, developers haven't really needed to alter code much for Intel. It's mainly been about clock speed.
Hopefully I'm wrong though!
Can't be if the game is not on console though.
Games like total war, hearts of iron 4, and anything on the clauswitz engine for that matter, seem to perform better on Intel,
Caporegime
- Joined
- 18 Sep 2009
- Posts
- 30,292
- Location
- Dormanstown.
You don't half make up some absolute twaddle.
Intel still has the advantage core for core, especially when it doesn't hit the HT/SMT threads.
There are no functional differences needing special programming.
It was Bulldozer which would have needed special programming to utilize it efficiently.
Pentium 4/Netburst, designed only for big marketing clock speeds, was pretty much similar screw up earlier.
Zen is AMD's return to more traditional core design, like what Intel's been using. (since their screw up)
Also Zen2 exactly adds features present in Intels which AMD couldn't fit (resources/time/die size...) into Zen1, like wider AVX unit.
Mesh/ring buses etc aren't really something programmers deal with.
Or registers/caches, whose improvements automatically improve performance without programmers needing to change anything in code.
Also would expect Zen2 to solve/mitigate problem of Zen(1)'s Infinity Fabric being hard locked into some fixed ratio with memory bus.
Which has made performance depend more on memory speed than with Intels.
I don't expect such a thing. I expect higher memory clocks up to 4000 to be supported and performance to scale accordingly.
no one is going to stone anyone .....even if they do say nvidia!
Blasphemer !!!!
Deleted member 209350
D
Deleted member 209350
https://www.youtube.com/watch?v=wjVjHce8uME
New AdoredTV video on Zen2! Basically confirms what we've already all seen, can't wait for zen2!
New AdoredTV video on Zen2! Basically confirms what we've already all seen, can't wait for zen2!
https://www.youtube.com/watch?v=wjVjHce8uME
New AdoredTV video on Zen2! Basically confirms what we've already all seen, can't wait for zen2!
Another half hour ramble that could be condensed into a quarter of the time, **** that.