Rocket Lake benchmark leaked

Nightfire

2[H]4U
Joined
Sep 7, 2017
Messages
3,068
Alot better? It shows the same SC score as a 5800x. The MC score is between that of a 10700k and a 5800x meaning HT is really bad or that all core frequencies are really low.

My guess is the latter based power consumption rumors. Still should be a good cpu based on other features like the igpu and quick sync.
 

Nightfire

2[H]4U
Joined
Sep 7, 2017
Messages
3,068
Clock speeds leaked show a 17% ipc increase using the above benchmark.

The 11900k has the same single core freq but 100 mhz less all core. I suspect this was to hit the 250 watt max tdp PL2 state. Very possible that it can reach 10900k clocks when overclocking - just make sure you fire up the reactor first.

Honestly though, higher clocks than I expected from RKL.
 

LukeTbk

Gawd
Joined
Sep 10, 2020
Messages
1,014
The MC/SC ratio

10700K: 6.655
5800x: 6.2414
11xxxK: 5.947

Is that a sign those company get better at boosting a single core when needed ?

Does seem nice but feel a step behind, an AMD refresh a la XT could let them kept the crowd in all scenarios if those numbers are a good indication, than again 10900K was closer in gaming performance than those single thread gap are on that geekbench, so maybe not.
 

Nightfire

2[H]4U
Joined
Sep 7, 2017
Messages
3,068
The MC/SC ratio

10700K: 6.655
5800x: 6.2414
11xxxK: 5.947

Is that a sign those company get better at boosting a single core when needed ?

Does seem nice but feel a step behind, an AMD refresh a la XT could let them kept the crowd in all scenarios if those numbers are a good indication, than again 10900K was closer in gaming performance than those single thread gap are on that geekbench, so maybe not.

You need more data. It could mean SMT performance is getting worse or there is a bigger discrepancy in clocks between SC and MC so as to keep PL2 down (which is the case for RKL).
 

Nightfire

2[H]4U
Joined
Sep 7, 2017
Messages
3,068

sram

[H]ard|Gawd
Joined
Jul 30, 2007
Messages
1,391
Clock speeds leaked show a 17% ipc increase using the above benchmark.

The 11900k has the same single core freq but 100 mhz less all core. I suspect this was to hit the 250 watt max tdp PL2 state. Very possible that it can reach 10900k clocks when overclocking - just make sure you fire up the reactor first.

Honestly though, higher clocks than I expected from RKL.
17% IPC increase is good considering they are in the same process.
 

Kdawg

[H]ard|Gawd
Joined
Aug 12, 2017
Messages
1,039
if intel beat amd to the 7nm, amd would be slightly behind right?
 

Nightfire

2[H]4U
Joined
Sep 7, 2017
Messages
3,068
i wish they could test power consumption clock for clock too.

i'm curious which uses more power at 4ghz

Considering the 11900k has similar pl2 as the 10900k (as leaked), they should be similar at 4 ghz too.
 

AzixTGO

n00b
Joined
Feb 21, 2016
Messages
25
Node process doesn't affect IPC
why not? process can affect transistor performance, can't it?

though in this case it might just be changes in the backporting process that causes a loss in IPC compared to what it would be if the chip was made on its intended node.
 

kirbyrj

Fully [H]
Joined
Feb 1, 2005
Messages
27,645
why not? process can affect transistor performance, can't it?

though in this case it might just be changes in the backporting process that causes a loss in IPC compared to what it would be if the chip was made on its intended node.

Indirectly maybe as in a smaller node would theoretically require less power and less heat is produced allowing for higher clocks which would increase performance.

Directly, no. The best comparison I can think of off the top of my head is older consoles (think XB360/PS3). Initially they were larger nodes and closer to the end of the production cycle, they were smaller nodes which yielded essentially the same performance. They just ran cooler and with less heat produced (and then you also had "slim" versions, etc. since there was less heat).
 

Nightfire

2[H]4U
Joined
Sep 7, 2017
Messages
3,068
why not? process can affect transistor performance, can't it?

though in this case it might just be changes in the backporting process that causes a loss in IPC compared to what it would be if the chip was made on its intended node.
Given all else the same, transistor size will affect maximum frequency and efficiency, but it doesn't really affect instructions PER CLOCK or IPC.
 

vick1000

2[H]4U
Joined
Sep 15, 2007
Messages
2,163
They could with hand-tuning of the design, but why bother? Not worth the investment.
Why bother? IDK, to match or surpass Intel? My point is they are clocking the chips as far as they can, just like Intel, it's meaningless to bring up IPC per clock. Overall performance in your chosen application and useage scenario, is what matters.
 

BrotherMichigan

Limp Gawd
Joined
Apr 8, 2016
Messages
349
Why bother? IDK, to match or surpass Intel? My point is they are clocking the chips as far as they can, just like Intel, it's meaningless to bring up IPC per clock. Overall performance in your chosen application and useage scenario, is what matters.
"IPC per clock."

They're already matching or surpassing Intel and Rocket Lake most likely isn't going to be worth buying even if they claw back a 5-10% single threaded performance lead. That performance lead will shrink as you pile on the threads, and it will run out of threads long before AMD, all while consuming around twice the power.
 

vick1000

2[H]4U
Joined
Sep 15, 2007
Messages
2,163
"IPC per clock."

They're already matching or surpassing Intel and Rocket Lake most likely isn't going to be worth buying even if they claw back a 5-10% single threaded performance lead. That performance lead will shrink as you pile on the threads, and it will run out of threads long before AMD, all while consuming around twice the power.
You brought up the fact that "Intel just clocks higher", that AMD has better IPC. That's why I responded with AMD would clock higher if they could, because they cannot. Should I underclock my Intel rig for some reason, just because AMD can't match my clocks?

I don't care, and no one should, about clock speeds or IPC. It's all about what performance I get in the end for my usage scenario and price point. If you need better multi-thread performance, it's the same arguement, IPC and clock speed are irrelevant, only the final performance that fits your needs matters. For games, Intel is still a good value at every tier, and ha a better reputation for stability and compatability. If you can use more than eight cores and sixteen threads, or course a stable AMD rig would make sense, as long as it fits your budget, since AMD is no longer the budget king in every tier.
 

BrotherMichigan

Limp Gawd
Joined
Apr 8, 2016
Messages
349
You brought up the fact that "Intel just clocks higher", that AMD has better IPC.
I didn't actually. Maybe you're thinking of someone else?
I don't care, and no one should, about clock speeds or IPC. It's all about what performance I get in the end for my usage scenario and price point.
This I agree with, and it's something Intel has a problem with currently (and will until Alder Lake is released.)
For games, Intel is still a good value at every tier, and ha a better reputation for stability and compatability.
This ranges from debatable to outright FUD.
 

MrGuvernment

Fully [H]
Joined
Aug 3, 2004
Messages
19,874
Is there going to be a 10 core or 12 core release under a different name at all? Or is Intel just doing 8 core and then moving off of 14nm ? (I am in the worst time to be wanting to buy a new cpu!)
 

Kdawg

[H]ard|Gawd
Joined
Aug 12, 2017
Messages
1,039
"IPC per clock."

They're already matching or surpassing Intel and Rocket Lake most likely isn't going to be worth buying even if they claw back a 5-10% single threaded performance lead. That performance lead will shrink as you pile on the threads, and it will run out of threads long before AMD, all while consuming around twice the power.


i want reviewers to look at IPC per watt.

don't care about clocks if the power draw is different.
 

kirbyrj

Fully [H]
Joined
Feb 1, 2005
Messages
27,645
I didn't actually. Maybe you're thinking of someone else?

This I agree with, and it's something Intel has a problem with currently (and will until Alder Lake is released.)

This ranges from debatable to outright FUD.

I initially brought it up posting that 4Ghz locked review showing AMD with a 8% IPC lead at the same clockspeed.

Performance is important, but not all desktops are equal. A SFF system might not have the cooling capacity to handle an all core OC'd i9.

AMD is just as stable as Intel, and the compatibility argument had more to do with the early Zen1 design. I haven't had a memory compatibility problem since my first 1700/B350 build 5 years ago.

i want reviewers to look at IPC per watt.

don't care about clocks if the power draw is different.

Definitely a talking point. I'm interested to see what a RKL 6C/12T CPU looks like (a 11400, not a 11600k). Especially since AMD starts at $299 right now. I'm guessing AMD will match Intel's pricing though and probably release a 5600 (non-x).
 

kirbyrj

Fully [H]
Joined
Feb 1, 2005
Messages
27,645
Is there going to be a 10 core or 12 core release under a different name at all? Or is Intel just doing 8 core and then moving off of 14nm ? (I am in the worst time to be wanting to buy a new cpu!)

Maybe something in their HEDT lineup? My understanding is that 8C16T is going to cap out their mainstream lineup.
 

MrGuvernment

Fully [H]
Joined
Aug 3, 2004
Messages
19,874
Thats a shame, the 10900k / 10850 were looking tempting for me since i can not find AMD anywhere! but i need cores!

So could be the ice-lake then but that takes us into Xeon line...but that doesnt look promising
Ice Lake-X, the HEDT lineup based on the new 10nm Sunny Cove core architecture is also nowhere in sight. Either it has been scrapped or delayed to 2021 as well.
https://www.hardwaretimes.com/intel...-lake-x-hedt-cpus-reportedly-delayed-to-2021/

Just trying to get caught up on Intel lately due to AMD shortage, i see LGA1200 is dead as of Alder Lake which could ship end of 2021....
 
Last edited:

Kdawg

[H]ard|Gawd
Joined
Aug 12, 2017
Messages
1,039
damn craptastic integrated graphics. only 50% improvement at the top end,
25% improvement on the core i5

11thgendes_575px.png
 

Comixbooks

[H]F Junkie
Joined
Jun 7, 2008
Messages
16,346
I would pick one up but ebay would be the only route for the next 9 months.
 

OrangeKhrush

[H]ard|Gawd
Joined
Dec 15, 2016
Messages
1,641
I initially brought it up posting that 4Ghz locked review showing AMD with a 8% IPC lead at the same clockspeed.

Performance is important, but not all desktops are equal. A SFF system might not have the cooling capacity to handle an all core OC'd i9.

AMD is just as stable as Intel, and the compatibility argument had more to do with the early Zen1 design. I haven't had a memory compatibility problem since my first 1700/B350 build 5 years ago.



Definitely a talking point. I'm interested to see what a RKL 6C/12T CPU looks like (a 11400, not a 11600k). Especially since AMD starts at $299 right now. I'm guessing AMD will match Intel's pricing though and probably release a 5600 (non-x).
missing a lot from AMD still

5100
5200 APU
5300
5400 APU
5500
5600
Threadrippers

The 11400 looks meh a $200 5600 and a 5300 around $160 should keep things in check
 

kirbyrj

Fully [H]
Joined
Feb 1, 2005
Messages
27,645
missing a lot from AMD still

5100
5200 APU
5300
5400 APU
5500
5600
Threadrippers

The 11400 looks meh a $200 5600 and a 5300 around $160 should keep things in check

Well, sure, but as it stands now, it looks like AMD will be ceding the sub-$300 market to Intel unless AMD decides to stealth launch something. For better or worse, they are victims of their own success. They launched the new consoles, new CPUs, and new graphics cards all in the same month. Supply is tight. If I were them, I'd be pumping out the high margin parts before the low margin ones.
 
Top