• 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.

Alder Lake-S leaks

Associate
Joined
13 Jan 2018
Posts
1,221
I'll have a quick scout around for some of these benchmarks to see how they actually compare between Rocket Lake and Zen 3
According to gamers nexus 11900k was pipped by the 5900x and 5950x in hitman 3
It total war three kingdoms the 11900k was faster than 5950x by 0.2% in average fps
 
Associate
Joined
29 Jan 2015
Posts
357
Yeah they are true, have a look at Age of Empire, Troy: A Total War Saga and Mount and Blade II they are all the same sort of game with huge AI simulation, its all multi threaded Integer math, its more like Microsoft Excel than a game, the slide seems oddly padded out with games like that, even Hitman 3 is to some extent like that.

What's left after that, Farcry 6, yeah ok tho we know Ryzen doesn't play well, again, specifically with Duna Engine.

Grid 2019, yeah sure, a bit specific again tho isn't it?

F1 2021, yeah ok.

That's it....

And yes there are questions at to the Windows 11 situation, if this is pre Ryzen fix?

I'll have a quick scout around for some of these benchmarks to see how they actually compare between Rocket Lake and Zen 3, i'm probably not going to find anything useful if at all as reviewers seem oddly reluctant to include CPU benchmarks when testing games these days or if they do its all on GPU limits.

For the strategy games the FPS is not really the most important factor. 30fps or 120fps in civ 6 makes very little difference but what does matter more is turn time. In Stellaris (I expect ADL to be really strong in this game) 60 fps is all you need but what matters late game is tic rate. FPS for strategy games is the wrong way to test them IMO.

The game suite does not seem all that comparable to most gaming test suites others use. TPU is a good place to go for CPU tests because they do test at 720p which is more CPU bound their current CPU gaming suite is BFV, Borderlands 3, CIV 6 FPS (pointless), CS:GO, CP2077, Doom Eternal, Far Cry 5 (expect this to be 6 for ADL / Zen 3d), Metro Exodus, RDR2 and Shadow of the Tomb Raider.

Only one game in common with the games Intel used, maybe 2 if they update their suite to include FC6 instead of FC5 and that game shows a performance loss to Zen 3 (although 3% is close enough to call it a tie).

HUB/Techspot use F1 2020, Rainbow Six Siege, Horizon Zero Dawn, Borderlands 3, Watch Dogs: Legion, Death Stranding, Shadow of the Tomb Raider and Hitman 2 so again only 1 overlapping game with the Intel test suite. Unfortunately they test at 1080p Max settings so it is more GPU bound than the TPU 720p tests.

Anandtech have a bit of an odd suite but they use Chernobylite, Civ 6, Deus Ex MD, FF14, FF15, World of Tanks, Borderlands 3, F1 2019, Gears Tactics, GTA V, RDR2 and Strange Brigade in DX 12 and Vulkan APIs.

For the sites that actually test 720p and lower where you are more CPU limited then I can see very different conclusions to the sites that start testing at 1080p Max and we will end up in the 720p is not real world so why test it nonsense even though it is an indicator of how the CPU will perform when you upgrade your GPU and alleviate the GPU bottleneck at 1080p.
 
Caporegime
Joined
17 Mar 2012
Posts
47,380
Location
ARC-L1, Stanton System
For the strategy games the FPS is not really the most important factor. 30fps or 120fps in civ 6 makes very little difference but what does matter more is turn time. In Stellaris (I expect ADL to be really strong in this game) 60 fps is all you need but what matters late game is tic rate. FPS for strategy games is the wrong way to test them IMO.

The game suite does not seem all that comparable to most gaming test suites others use. TPU is a good place to go for CPU tests because they do test at 720p which is more CPU bound their current CPU gaming suite is BFV, Borderlands 3, CIV 6 FPS (pointless), CS:GO, CP2077, Doom Eternal, Far Cry 5 (expect this to be 6 for ADL / Zen 3d), Metro Exodus, RDR2 and Shadow of the Tomb Raider.

Only one game in common with the games Intel used, maybe 2 if they update their suite to include FC6 instead of FC5 and that game shows a performance loss to Zen 3 (although 3% is close enough to call it a tie).

HUB/Techspot use F1 2020, Rainbow Six Siege, Horizon Zero Dawn, Borderlands 3, Watch Dogs: Legion, Death Stranding, Shadow of the Tomb Raider and Hitman 2 so again only 1 overlapping game with the Intel test suite. Unfortunately they test at 1080p Max settings so it is more GPU bound than the TPU 720p tests.

Anandtech have a bit of an odd suite but they use Chernobylite, Civ 6, Deus Ex MD, FF14, FF15, World of Tanks, Borderlands 3, F1 2019, Gears Tactics, GTA V, RDR2 and Strange Brigade in DX 12 and Vulkan APIs.

For the sites that actually test 720p and lower where you are more CPU limited then I can see very different conclusions to the sites that start testing at 1080p Max and we will end up in the 720p is not real world so why test it nonsense even though it is an indicator of how the CPU will perform when you upgrade your GPU and alleviate the GPU bottleneck at 1080p.

Yeah, testing with maximum IQ settings is ok, in fact it is what i do because your IQ settings can have a dramatic impact on the CPU, it can be the difference between loading a couple of threads or 16.

You just turn the resolution down if what you're trying to get to is the performance of the CPU, even if that is 480P, if the GPU is the limit what you're testing is the GPU, not the CPU, you can't then call it a "CPU game benchmark" it isn't, looking at you Hardware Unboxed.
 
Associate
Joined
29 Jan 2015
Posts
357
Yeah, testing with maximum IQ settings is ok, in fact it is what i do because your IQ settings can have a dramatic impact on the CPU, it can be the difference between loading a couple of threads or 16.

You just turn the resolution down if what you're trying to get to is the performance of the CPU, even if that is 480P, if the GPU is the limit what you're testing is the GPU, not the CPU, you can't then call it a "CPU game benchmark" it isn't, looking at you Hardware Unboxed.

Yea that is why I think TPU are probably the best to use since they test 720P Max as well as other resolutions. If Anandtech added lowest res + Max IQ to their test suite (and since Ian automated most of it I don't think this is hard for him to do but what do I know?) they would also be a good goto for pure stock metrics since they use Jedec spec timings + Intel spec IMC speed + Intel spec power and tau.
 
Caporegime
Joined
17 Mar 2012
Posts
47,380
Location
ARC-L1, Stanton System
Yea that is why I think TPU are probably the best to use since they test 720P Max as well as other resolutions. If Anandtech added lowest res + Max IQ to their test suite (and since Ian automated most of it I don't think this is hard for him to do but what do I know?) they would also be a good goto for pure stock metrics since they use Jedec spec timings + Intel spec IMC speed + Intel spec power and tau.

What do you know? more than most reviewers....
 
Permabanned
Joined
2 Sep 2017
Posts
10,490
Yea that is why I think TPU are probably the best to use since they test 720P Max as well as other resolutions. If Anandtech added lowest res + Max IQ to their test suite (and since Ian automated most of it I don't think this is hard for him to do but what do I know?) they would also be a good goto for pure stock metrics since they use Jedec spec timings + Intel spec IMC speed + Intel spec power and tau.

Low resolution games test is not proper gaming experience test.
The reviewer must stress on typical gaming settings, for everything else there are software which use only the CPU.
 
Soldato
Joined
6 Oct 2007
Posts
22,260
Location
North West
Intel-Alder-Lake-Embargo.jpg
 
Associate
Joined
29 Jan 2015
Posts
357
Low resolution games test is not proper gaming experience test.
The reviewer must stress on typical gaming settings, for everything else there are software which use only the CPU.

It is not meant to be a current gaming experience test. It is meant to simulate a future gaming experience test when you pair up that CPU with a more powerful GPU.

So you take your old 2700K + GTX 580 and run it at a low res + max IQ to simulate how a 2700K + GTX 980 will perform at your target res + max IQ. For 4k gamers that might mean using the 1080p results to guess at future performance as the bottleneck goes from GPU to CPU. For 1080p high refresh rate gamers you want sub 1080p resolutions to simulate that.

Obviously if you intend to upgrade your whole platform every 3 years then all you really care about is current performance at your target res which is fine but for those who want to hang onto a platform for 6+ years it has predictive value so TPU doing 720p upto 4k is covering everybody since those who care about predicting longer future performance have data they can use and those who only care about the short term are also catered to.
 
Permabanned
Joined
2 Sep 2017
Posts
10,490
It is not meant to be a current gaming experience test. It is meant to simulate a future gaming experience test when you pair up that CPU with a more powerful GPU.

So you take your old 2700K + GTX 580 and run it at a low res + max IQ to simulate how a 2700K + GTX 980 will perform at your target res + max IQ. For 4k gamers that might mean using the 1080p results to guess at future performance as the bottleneck goes from GPU to CPU. For 1080p high refresh rate gamers you want sub 1080p resolutions to simulate that.

Obviously if you intend to upgrade your whole platform every 3 years then all you really care about is current performance at your target res which is fine but for those who want to hang onto a platform for 6+ years it has predictive value so TPU doing 720p upto 4k is covering everybody since those who care about predicting longer future performance have data they can use and those who only care about the short term are also catered to.

This depends on the games' readiness to work with the CPU in the test. Because if it utilises only 14% under 3840x2160, then obviously that particular game is not properly optimised.
There is no way to simulate future proof via lowering the resolution alone.
 
Soldato
Joined
15 Oct 2019
Posts
11,656
Location
Uk
Your not going to notice much difference in the future anyway, take for instance pairing a 4790k or a 6700k with a 3090 @ 4k the difference would be hardly noticeable.
 
Associate
Joined
11 Dec 2016
Posts
2,010
Location
Oxford
If it had CS:GO and Rocket League +15% on those slides i would be impressed
you're being too suspicious here. Yes, those and Dota, LoL are the best gaming IPC markers.
But they are showing an actual slower and equal results there.

From those results, I would pick F1 as representative.
Valid +8% looks like a win. And a masive improvement over their own previous gen.

Except its the 5.2? GHz top end SKU 12900K. You can expect 5800X to match 5950X in most games. Good luck getting same mileage out of 12600K and 12700K, gimped by smaller cache and lower boost

Has anyone seen anything on how Linux will handle the big/little core architecture ?
Everything I've seen so far only seems to talk about Windows (usually 11).
From a quick glance in phoronix, ther have been linux kernel patches for Alder Lake power management and integrated graphics. But nothing about thread director or big.little
 
Caporegime
Joined
17 Mar 2012
Posts
47,380
Location
ARC-L1, Stanton System
4690K + GTX 970 (perfectly good)
4690K + GTX 1070 (couldn't cope with it, very high CPU loads, GPU bottlenecks even at 1440P and Microstutter even in old games like Insurgency.

Upgraded the CPU to a Ryzen 1600. Cured all the problems, GTX 1070 was stretching its legs like Usain Bolt....

Star Citizen got updated with new Landing Zones and the 1600 was struggling with them, upgraded to a Ryzen 3600, much better.

Upgraded to an RTX 2070 Super, CPU was starting to become a problem in Star Citizen again, tho it was not bad, certainly nothing like the problems i had with the 4690K and the GTX 1070.

Upgraded the CPU to a Ryzen 5800X in anticipation of an RX 6800 that's probably never going to happen now... Star Citizen runs great tho.
--------------

You know what i hate, when people doggedly are insisting what they think is true trumps everyone else's opinion, that AMD and Intel should only make 6 core £200 CPU's or if they do they should only cost a few quid more, because F people who want something better than that?
 
Associate
Joined
25 Apr 2017
Posts
1,095
Is the top-end Zen 3D-V cache 6950X expected to have a lower TDP than the 12900k? Only interested in a new CPU for gaming so DDR5 and PCIE 5.0 doesn't really interest me the boards and RAM would cost a fortune.
 
Permabanned
Joined
2 Sep 2017
Posts
10,490
Is the top-end Zen 3D-V cache 6950X expected to have a lower TDP than the 12900k? Only interested in a new CPU for gaming so DDR5 and PCIE 5.0 doesn't really interest me the boards and RAM would cost a fortune.

Now the choice is between Alder Lake and Zen 4 whenever it comes, maybe late 2022. Who knows... :o
 
Back
Top Bottom