• 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.

Raptor Lake Leaks + Intel 4 developments

Caporegime
Joined
17 Mar 2012
Posts
47,669
Location
ARC-L1, Stanton System
I suggest reading more on what features Raptor Lake is said to bring, such as this:

Intel Raptor Lake CPUs To Feature Digital Linear Voltage Regulator ‘DLVR’ – Could Help Reduce Power Consumption By 25%​



Though of course it's obvious that a 24core CPU (lets assume similar IPC/frequency, though it's likely RPL with be higher than Zen4 on both) will consume more power than a 16 core CPU in MT workloads, and deliver much higher performance in such workloads.


Are you sure about that? the demo put the 7950X 46% ahead of the 12900K in MT, that's an R23 score of over 40,000 for the 7950X, the 12900K scores 27,472, the half cores are only worth 9,000, that's 36,472 for the 13900K
 
Soldato
Joined
31 Oct 2002
Posts
9,865
Are you sure about that? the demo put the 7950X 46% ahead of the 12900K in MT, that's an R23 score of over 40,000 for the 7950X, the 12900K scores 27,472, the half cores are only worth 9,000, that's 36,472 for the 13900K

Check what planet/thread your on bro, discussing the 13900k (24 cores), not the 12900K or 12900KS.
 
Caporegime
Joined
17 Mar 2012
Posts
47,669
Location
ARC-L1, Stanton System
@Dave2150

Zen 4 16 core blender demo, completed 30 something % before the 12900K, the agreed maths is the 12900K would have to be 46% faster to match the 16 core Zen 4

12900K R23 scores 27,472 + 46% = 40,109, that's what the 16 core Zen 4 should score in R23, the half cores are worth + 9,000 points in R23 = 36,472 for the 13900K
 
Soldato
Joined
19 Jan 2022
Posts
2,753
Location
Devilarium
@Dave2150

Zen 4 16 core blender demo, completed 30 something % before the 12900K, the agreed maths is the 12900K would have to be 46% faster to match the 16 core Zen 4

12900K R23 scores 27,472 + 46% = 40,109, that's what the 16 core Zen 4 should score in R23, the half cores are worth + 9,000 points in R23 = 36,472 for the 13900K
Not really. In most blender stuff the 12900k loses to the 5950x, so using blender data to CBR23 is flawed. The only blender test the 12900k is doing well is the BMW.
 
Caporegime
Joined
17 Mar 2012
Posts
47,669
Location
ARC-L1, Stanton System
Not really. In most blender stuff the 12900k loses to the 5950x, so using blender data to CBR23 is flawed. The only blender test the 12900k is doing well is the BMW.

Its depends, See Linus, he said in shorter runs the 12900K actually beats the 5950X in Blender, because the 12900K boosts much higher initially and then gets heat soaked and lowers clocks, the Blender demo was under 4 minutes.
-----------------

Because of the 140 watt PGA socket limit the 5950X actually only runs at about 4Ghz in Blender / R23, some a little higher, some a little lower, depends on the silicon.

AM5 has a much higher socket limit, i think Robert said about 250 watts with a 170 watt CPU TDP, without wanting to give much away he said the demo CPU was running at more than 105 watts but less than 170 watts, the thinking is Zen 4, even the 16 core variant would run at over 5Ghz all core in R23 / Blender, reportedly an engineering sample was running at 5.8Ghz in lighter workloads, 5.2Ghz would be 30% higher than the 5950X, the rest could be down to IPC. its actually only 35% to 40% faster than the 5950X.
 
Soldato
Joined
22 May 2010
Posts
11,896
Location
Minibotpc
So once again it seems Intel are looking to have Raptor Lake being the last gen on what I consider a brand new socket, do they just like creating e-waste or something? Why will/would anyone buy a new Z790 board, instead of the cheaper Z690/B660 options now? What else can they add, or maybe they'll try to compete with AMD and have boards with full PCI-E 5.0 on all devices?
What a waste of time, this is why it makes AMD more interesting in the long run with the support they give on multi generation of chips supported by a single generation of socket across multiple chipsets etc.
 
Soldato
Joined
19 Jan 2022
Posts
2,753
Location
Devilarium
Its depends, See Linus, he said in shorter runs the 12900K actually beats the 5950X in Blender, because the 12900K boosts much higher initially and then gets heat soaked and lowers clocks, the Blender demo was under 4 minutes.
-----------------

Because of the 140 watt PGA socket limit the 5950X actually only runs at about 4Ghz in Blender / R23, some a little higher, some a little lower, depends on the silicon.

AM5 has a much higher socket limit, i think Robert said about 250 watts with a 170 watt CPU TDP, without wanting to give much away he said the demo CPU was running at more than 105 watts but less than 170 watts, the thinking is Zen 4, even the 16 core variant would run at over 5Ghz all core in R23 / Blender, reportedly an engineering sample was running at 5.8Ghz in lighter workloads, 5.2Ghz would be 30% higher than the 5950X, the rest could be down to IPC. its actually only 35% to 40% faster than the 5950X.
If Linus said that he is clueless. Unless he is thermal throttling, which makes it an invalid test anyways, the 12900k never drops clocks. It straight up loses in blender though, except the BMW test. On the other hand it also straight up wins in CBR23.
 
Caporegime
Joined
17 Mar 2012
Posts
47,669
Location
ARC-L1, Stanton System
If Linus said that he is clueless. Unless he is thermal throttling, which makes it an invalid test anyways, the 12900k never drops clocks. It straight up loses in blender though, except the BMW test. On the other hand it also straight up wins in CBR23.
Again does it? why aren't they topping the R23 thread?

The numbers i'm working from is CPU Monkey.

12900K: 27,472
5950X 28,722
 
Soldato
Joined
19 Jan 2022
Posts
2,753
Location
Devilarium
Again does it? why aren't they topping the R23 thread?

The numbers i'm working from is CPU Monkey.

12900K: 27,472
5950X 28,722
Those are not stock numbers for either of them. The 12900k gets around 27900-28k in high priority, and the 5950x around 25k to 25500. Again, we are talking about STOCK numbers, not PBO, not auto OC, not enhanced turbos etc.
 
Caporegime
Joined
17 Mar 2012
Posts
47,669
Location
ARC-L1, Stanton System
Those are not stock numbers for either of them. The 12900k gets around 27900-28k in high priority, and the 5950x around 25k to 25500. Again, we are talking about STOCK numbers, not PBO, not auto OC, not enhanced turbos etc.

Stock the 5950X varies a bit, as i explained, it depends on how many Mhz the silicon can mange with in its socket power limitations, i'm also willing to concede the CPU Monkey numbers are on the good silicon side, yet they are stock, especially for later silicon. But i think yours are on the bad silicon side.
 
Soldato
Joined
19 Jan 2022
Posts
2,753
Location
Devilarium
Stock the 5950X varies a bit, as i explained, it depends on how many Mhz the silicon can mange with in its socket power limitations, i'm also willing to concede the CPU Monkey numbers are on the good silicon side, yet they are stock, especially for later silicon. But i think yours are on the bad silicon side.
Every review I've seen running stock puts the 5950x at 25500. Never seen a stock review at 28k or anything like that. Techspot and techpowerup seem to agree with my numbers
 
Caporegime
Joined
17 Mar 2012
Posts
47,669
Location
ARC-L1, Stanton System
Every review I've seen running stock puts the 5950x at 25500. Never seen a stock review at 28k or anything like that. Techspot and techpowerup seem to agree with my numbers

Two reviews? do you have a wider sample than two? There are a lot more than 2 reviewers. if i look for wider samples will they all agree with that ^^^^ i have already cited one which does, what if i can get 3, or 4?
 
Soldato
Joined
19 Jan 2022
Posts
2,753
Location
Devilarium
Two reviews? do you have a wider sample than two? There are a lot more than 2 reviewers. if i look for wider samples will they all agree with that ^^^^ i have already cited one which does, what if i can get 3, or 4?
Yes, I do. Just checked anandtech / computerbasede and toms hardware. Only tom's hardware gets 28k but that is with PBO on, it also has numbers at stock and that's at 26k. So..can we stop this already? It's just a fact, 5950x loses to cbr 23. Holy crap dude

You haven't cited a single one, cpu monkey does not do reviews. They just upload scores from users
 
Last edited:
Soldato
Joined
26 Aug 2004
Posts
5,032
Location
South Wales
Looks possible Intel might have up to 10% or more IPC advantage with Raptor Lake, hoping reviewers test with DDR4 not just DDR5.

A discounted MSI z690-a could be a decent choice for it if the performance is there. I'd happily get a 13700k if it's worth it.

Waiting on perf numbers.
 
Soldato
Joined
22 May 2010
Posts
11,896
Location
Minibotpc
Looks possible Intel might have up to 10% or more IPC advantage with Raptor Lake, hoping reviewers test with DDR4 not just DDR5.

A discounted MSI z690-a could be a decent choice for it if the performance is there. I'd happily get a 13700k if it's worth it.

Waiting on perf numbers.
I just wish there were more Z690I boards DDR4 available for the ITX market right now. It's so stale with the only options being Gigabyte Aorus Z690I which is currently borked due to that PCIE issue or the Asrock Z690I AX which is so anaemic, it shouldn't be called a Z board.
 
Associate
Joined
16 Feb 2020
Posts
88
CB R23
8P cores at 5ghz = 20700~ (164W) 1,15Vcore
4E cores at 4ghz = 4179~ (19W) 1,15Vcore
16E cores at 4ghz = 16 716~ (76W) 1,15Vcore

8P/16E 5ghz/4ghz 1,15Vcore 240W 37 416~
This is without any IPC and node improvements for RPL, if they can polish 10nm Super Fin to achieve 5ghzP 4ghz E at 1,1V or less it will be nice.
 
Back
Top Bottom