Arrow Lake vs Ryzen 9000

[Brexy9]

What do you guys think about the new Ryzen 9000 leaks and Intel Arrow Lake ones? I was really tempted to upgrade to Ryzen 7000 from my Ryzen 5 3600, but since Ryzen 9000 is coming out so soon I might upgrade to it or intel Ultra chips when they come out. I'm just skeptical of Intel not doing hyper-threading anymore and hope their Power Consuption goes down since all the issues the i7s and i9s had. What are your guys' thoughs on Ryzen 9000 vs Intel Ultra (based on the leaks we got) 

[GuiltySpark_]

I think its going to be the same as its always been. One slightly better than the other at this vs. that. 

 

If you're asking people to comment on "leaks", please provide the details. 

[podkall]

15 minutes ago, GuiltySpark_ said:

I think its going to be the same as its always been. One slightly better than the other at this vs. that. 

 

If you're asking people to comment on "leaks", please provide the details. 

I personally would start choosing after some basic benchmarks are out at least from the official products

[bal723]

25 minutes ago, Brexy9 said:

 I'm just skeptical of Intel not doing hyper-threading anymore and hope their Power Consuption goes down since all the issues the i7s and i9s had.

i am sure intel chips gonna have super low power consumptions by 100% as they are going to use 2nm nodes also hyper threading is not that important if you have alot of cores in first place, so basically core 3 should have 4P and 4E cores to compensate for it, i myself consider upgrading to lga 1851 but i am not sure if i can get anything in there for 300$ or under

[OddOod]

Intel might have solved their unholy power problem by dropping to a smaller process node, but it wouldn't surprise me if they only reel it in a small amount. My guess is that Intel is gonna keep their bare edge in "fastest gaming CPU" while AMD is gonna continue to *spank* them in performance per watt under load

 

34 minutes ago, bal723 said:

hyper threading is not that important if you have a lot of cores

mmmm, ish. SMT boosts per physical core performance by ~30-40% last I checked (which was a while ago so could be wrong) so it's pretty useful if you have all core workloads. Heck, Division 2 was sucking back 80% of my 32 threaded 5950X

[bal723]

3 minutes ago, OddOod said:

mmmm, ish. SMT boosts per physical core performance by ~30-40% last I checked (which was a while ago so could be wrong) so it's pretty useful if you have all core workloads. Heck, Division 2 was sucking back 80% of my 32 threaded 5950X

well, atleast thats what intel trying to say and i fully understand what they mean

[ki8aras]

1 hour ago, Brexy9 said:

What do you guys think about the new Ryzen 9000 leaks and Intel Arrow Lake ones? I was really tempted to upgrade to Ryzen 7000 from my Ryzen 5 3600, but since Ryzen 9000 is coming out so soon I might upgrade to it or intel Ultra chips when they come out. I'm just skeptical of Intel not doing hyper-threading anymore and hope their Power Consuption goes down since all the issues the i7s and i9s had. What are your guys' thoughs on Ryzen 9000 vs Intel Ultra (based on the leaks we got) 

 

45 minutes ago, bal723 said:

i am sure intel chips gonna have super low power consumptions by 100% as they are going to use 2nm nodes also hyper threading is not that important if you have alot of cores in first place, so basically core 3 should have 4P and 4E cores to compensate for it, i myself consider upgrading to lga 1851 but i am not sure if i can get anything in there for 300$ or under

early leaks are pointing to arrow lake showing a performance regression compared to raptor lake lake so i wouldnt count on anything too amazing happening

[bal723]

2 minutes ago, ki8aras said:

 

early leaks are pointing to arrow lake showing a performance regression compared to raptor lake lake so i wouldnt count on anything too amazing happening

still a good platform tbh, also the arc archeticture igpus may help me with deep link and streaming

[BiG StroOnZ]

5 hours ago, ki8aras said:

early leaks are pointing to arrow lake showing a performance regression compared to raptor lake lake so i wouldnt count on anything too amazing happening

 

Can you point me to this leak, these two claim 15-40% performance boost over Meteor Lake:

 

Arrow Lake could offer huge performance gains, but AMD’s Zen 5 might still defeat Intel’s mighty next-gen CPUs | TechRadar

Intel 15th-gen Arrow Lake: Everything we're expecting (xda-developers.com)

 

Then there's this leak that suggests 16-20% uplift over Raptor Lake in Geekbench:

 

Intel 15th Gen Arrow Lake: Everything we know - Dexerto

 

I know Raptor Lake Refresh has the core count advantage and HyperThreading, therefore, I'm looking forward to seeing how Intel combats this (to make-up for the loss in thread count). When compared to Raptor Lake Refresh, Meteor Lake doesn't do that bad, considering it is having less cores/threads usually. Then with Arrow Lake, they are also doing away with HT. Meaning, it will be interesting to find out how Intel expects to overcome that disparity in Benchmarks. Otherwise, it won't look great from reviewers if there is a regression in performance. 

 

If they can manage: lower TPD, better efficiency, better thermals, have lower thread counts, while also doing away with HT/SMT (yet still tie Raptor Lake Refresh). I think most people will be happy with that outcome. Especially, if gaming performance shows no regression. Only problem is even if AMD manages a meager 15-20% increase in performance from Zen4 to Zen5, like what is suggested from leaks, AMD will still take the performance crown. 

 

6 hours ago, bal723 said:

still a good platform tbh, also the arc archeticture igpus may help me with deep link and streaming

 

Definitely this. I'm also wondering if iGPU tech can get so good to the point where we can start using it in conjunction with dGPUs (for Hybrid GPU power, as seen with Lucid Virtu). 

[bal723]

11 hours ago, BiG StroOnZ said:

Definitely this. I'm also wondering if iGPU tech can get so good to the point where we can start using it in conjunction with dGPUs (for Hybrid GPU power, as seen with Lucid Virtu). 

they already have so called dynamic power sharing and deep link where both igpu and dgpu work together or put specific type of load on eachother

[jaslion]

On 5/24/2024 at 3:46 PM, OddOod said:

fastest gaming CPU" while AMD is gonna continue to *spank* them in performance per watt under load

I mean they dont even have that. You need a custom loop cooled 14900ks with unlimited power budget enabled with allowance for all core 6ghz to sometimes have it beat a 7800x3d.

 

Those are normally all not stock settings at all and even then it will struggle to keep an edge. Its silly at this point

[porina]

On 5/24/2024 at 2:46 PM, OddOod said:

mmmm, ish. SMT boosts per physical core performance by ~30-40% last I checked (which was a while ago so could be wrong) so it's pretty useful if you have all core workloads.

It depends a lot on the workload and the average will be far lower than that. That range might apply to Cinebench which is towards the best case for HT/SMT.

 

Also note that benefit is in overall throughput, not per-thread performance which actually goes down. Removing HT/SMT greatly reduces the chance of Windows doing something stupid in scheduling. In the worst case HT/SMT provides practically zero performance benefit while still significantly increasing power consumption.

 

 

As for the original question, we wont know until they're released and tested. We might get an early clue how Arrow Lake might perform as the earlier release low power optimised Lunar Lake will use the same P core architecture. If the rumours are true that AMD are doubling the FP execution resource in Zen 5 it'll be the first time for a very long time they'll have a clearly better FPU than Intel through the product stack.

[Wolverine2349]

Would love to see an Arrow Lake with more than 8 P cores on a single tile. But not going to happen per reports.

 

Would love to see AMD Zen 5 have more than 8 P cores on a single CCX on single CCD. Though appears going to be 8 core CCXs on one CCD like Zen 3 and 4. 

 

Hard to say how much improvement without seeing benchmarks.

 

Though some recent reports look underwhelming for Zen 5 compared to expectations before. Hearing 10% IPC uplift and similar clocks. Appears AMD with all this stupid obsessed AI hype is focusing too much on that. GPUs provide that and so sick of this race to put them on CPUs. The heck with AI IMHO. SO sick of that stupid bubble hype!!

 

Arrow Lake appears to be a small improvement because of clock speeds regression, but even so with hopefully better IMC and hopeful for better thermals and much lower power and most importantly super stable unlike 13th and 14th Gen which have degradation and random stability issues and no known power limits and even the PL1 and PL2 253W has stability issues in some cases..,

 

I hope Arrow Lake solves that.

 

And I really hope the IPC uplift in Zen 5 is at least 20%. We need more performance if we are still stuck at 8 cores per CCD for upcoming RTX 5090.

 

Otherwise staying with 7800X3D even when the 9000 X3D parts release of performance uplift is IPC wise more underwhelming than Zen 3 to 4. At least Zen 3 to 4 got 12% and a good clock speed bump which Zen 5 appears to not be getting. Shame on AMD for that and also rumored to drop WIN10 support because of this stupid AI garbage they want to put in CPUs!!

 

 

I would hope much higher IPC uplift than 10%. But all this focus on AI accelerators makes me worried that's all it will be.