• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Arrow Lake leaks + post-launch discussion

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
As was already said, I think they may release the 9800X3D earlier as it's the chip that will sell the best, and so far, the Ryzen 9000 series is selling much worse than expected. Even though these CPUs are good, most potential users are disappointed as they were expecting something better. Gamers wait for the X3D series, and if AMD won't make a move soon, then many of them will go the Intel way. The new AMD motherboards are also good, but the same, are not expected to sell well because they're about the same as the older gen. Those who already have B650/X670 probably won't see the point of the "upgrade". On the other hand, those who have earlier AMD motherboards and ARL convinces them will also get new Intel motherboards that will bring more new things like CUDIMM support, that probably barely anyone will use, but well, it's new and "future-proof" for many. So, new motherboards can be another reason why people move to Intel.

I'm not sure if marketing is so much better in pushing lies or if people are dumber in general, but for years, for gaming, CPUs don't matter much as long as in use is something quite modern with enough cores (typically everything with 6+ cores). Still, many people jump onto every new CPU generation as if it were a game changer. The only exception is a few titles people play at low display resolution for 300+ FPS. We were talking about it so many times that for many OCF members it's already obvious.

I assume that Intel will release its KS chip(s?) just after the whole X3D line is in stores. They do that each generation. Releasing the 9800X3D earlier will be a surprising move from the AMD's side, but also, for once, it will be reasonable. They usually release graphics cards or CPUs before Nvidia or Intel. Intel makes mistakes more often and has difficulty beating Ryzen, but Nvidia generally beats AMD each time. If AMD really releases the 9800X3D sometime soon, it will prove that AMD is afraid of ARL.
 
Intel slides have leaked. ARL seems to be a bit disappointing, especially for gaming (<=14900k post bios updates). Power consumption is significantly down at least. Keep in mind that these are Intel supplied results and that Intel is using APO for their gaming results which is a manual tuning effort by Intel and only supported in a very limited number of games. I expect independent reviews will show even worse gaming performance.

1728398655686.jpeg

1728398748343.jpeg
1728398788329.jpeg
1728398817816.jpeg
1728398840762.jpeg
 
Intel slides have leaked. ARL seems to be a bit disappointing, especially for gaming (<=14900k post bios updates). Power consumption is significantly down at least. Keep in mind that these are Intel supplied results and that Intel is using APO for their gaming results which is a manual tuning effort by Intel and only supported in a very limited number of games. I expect independent reviews will show even worse gaming performance.
Press briefing has happened ahead of public announcement and not surprising it gets leaked.

Finally had time to look a bit more and have an initial think about it. It's... kinda ok. Not wow. I guess in a similar way Zen 5 wasn't "wow" for anyone except those who like AVX-512, like myself. Skimming comment sections elsewhere, a common theme I do agree with is that given the major process node gain on Intel's side here, kinda hoping for a bit more. For what its worth they claim ST win. Otherwise this was about gaming and we're fishing in the depths for any differences. If AMD drop the 9800X3D it'll make that moot anyway. No MT productivity tests. The power claim I find a little odd. Why not give direct CPU readings instead of hiding it in a system result?

We're going to have to wait for more details. Let's see what the public announcement gives us in a couple days. Then 2 more weeks for review embargo.

Edit:
1728407944140.png
Found this one elsewhere. Slide is framed as CPUs capable of both gaming and work. Claiming a win vs the previous gen 7950X3D, so for pure work the non-X3D or Zen 5 offerings will be a lot closer.
 
Last edited:
Videocardz have leaked the English language slides ahead of the 8AM Pacific embargo (about 6 hours to go at time of writing). Looks to be same slides but easier to read without having to translate it.

Thinking more about the system power thing, I have to wonder if they did that to make reductions sound bigger, because I'd take that to be wall power which would include PSU loss. It's still realistic in the sense people pay for wall power, but for technical comparisons I'd prefer to see CPU socket power, with some risk they might not be reported exactly the same way between manufacturers which might also contribute to using system power.

I'm seeing some rumbles elsewhere, not from known or trusted sources to me, that gaming perf might be underwhelming. We saw what that did to Zen 5. I doubt we'll get any more info for another 2 weeks unless there are more leaks.
 
"The new Intel CPUs are Impressive and Disappointing at the same time"

0:00 Intro
0:38 Seasonic (Advertising)
1:07 Overview: Intel Core Ultra 200
4:30 Faster E-Cores
5:01 The Structure of the CPUs
6:10 Focus on Efficiency
7:57 Overclocking with Arrow Lake
9:35 Improved RAM Controller
10:52 PCIe Lanes
11:59 285K vs 13900K: CPU Comparison
15:03 New Motherboards, New Socket
16:48 ILM vs RL-ILM vs CF
18:00 Changes to the Socket & New Contact Frame
20:37 Problems for Reviewers?
21:44 Prices
22:20 Summary/Conclusion
23:27 Outro

 
Seems like a "fixed" 14900k that doesn't need water cooling. At least that's what I hope. Gpu is what needs the most power for gud grafx anyways. If this leaves more power for the gpu while still being pretty damn fast then it seems like a good direction to start down. Still, the psychological effect of being at least a little better than the last gen in games would have been huge. I want to see a review with a 265k with fast ddr5 vs a 12700k with ddr4 to see if it's worth it. Also not ruling out the 9800x3d.
 
Hallock did say they expect Arrow Lake to be around 5% behind current X3D in gaming. If 9000X3D can improve on that, it'll widen the gap. AMD will retain gaming leadership and gain some benefit from halo effect. However, that only applies for the top of the stack. Lower down, people are more sensitive to price vs performance and we'll have to see how it goes there.

It could be argued if 7800X3D or 7950X3D is the current ultimate performance gaming CPU, but 7800X3D will be the one most buy due to its lower price. We have existing Zen 5 pricing, and Intel has shown their Arrow Lake MSRPs. It'll be up to AMD to decide how it wants to position the 9800X3D when the time comes.

Leak has come out over the last day about "9000X3D". Vs 7800X3D in 1080p gaming with 4090: +2% each of BMW and SOTTR. 13% in Far Cry 6. We'll have to wait for benchmarkers to test much wider games for a better picture. No difference in Cinebench R23 "IPC" which I'd expect as that isn't cache/memory sensitive load. Overall MT does go up relative 7000 but again we saw that with non-3D parts already. Need to check if the uplift is similar, score can be used to work back to running clock at the time.

Edit: I done an initial estimate on the 9000X3D CB R23 MT clocks. 5.1+ GHz for 8 core, 4.6+ GHz for 16 core. Now to see if anyone mentioned running clocks on that test for current released (non-3D) models.
Edit2: Looking at their IPC results again, their efficiency seems to be about 4% off HWBOT results, unless I'm missing something. If we use that instead, 5.3+ GHz for 8 core, 4.8+ GHz for 16 core.
For comparison, my stock (+EXPO) 7800X3D runs that at 4.7 GHz+. My score is about 5% below theirs, but I'm running it with a ton of open stuff including monitoring for a quick look.
 
Last edited:
Finally watched der8auer's video over breakfast. Well, a part of it. Got to the ram part then I think he misspoke or otherwise was confused and I stopped there before I got any more obviously bad information. Wasn't great anyway since he was just regurgitating the slides so no useful additional info I could tell.

The only small benefit I got from having it on was that I now see ECC support on CPU being mentioned on the slide. Unfortunately I think this will like AMD situation. It wont be mandatory and I'm going to guess consumer/enthusiast mobos will NOT support it. I had a quick look at a high end Asus and Gigabyte boards and both do not support full ECC.
 
The intel website says the 265k uses more power than the 12700k so there goes that theory. I guess I'll just wait for reviews and hope for the best. If the newest gen cpus from amd and intel are both minimal upgrades this time around I guess I'll just limp along until the next series with my 12700k and 4090 lol.
 
As long as the PC isn't really old, CPUs were never a good reason to upgrade the whole PC. It's only that marketing pushes the whole "you need it" and people often spend money on pointless upgrades. Each generation gives +5-15% IPC. If your CPU is 1-2 generations back, then there is barely any difference in daily usage or even gaming (I'm not saying about something like P4 to C2D jump).
The main difference is in motherboards and additional features. If you need USB4, WiFi 7, faster LAN, faster or larger capacity RAM support, or more M.2 SSD support, then it can be a good idea to upgrade. However, most new things can be bought as separate PCIe cards. New motherboards are also very expensive, which makes people think twice about whether they need a new PC.

The 12700k is three generations older than the 265k, and TDP doesn't tell the real maximum wattage, so I guess we have to wait for detailed tests. There will be significant differences in performance and power management. I'm not saying it will be worth upgrading, but I wouldn't focus on the maximum power when the CPU runs at 10-30% most of the time. You may care about power usage if you leave it running 24/7 at a significant load. My 24/7 gaming PC uses a Ryzen 9 7945HX at 55W TDP. I have no problems with games at 1440p, and it's silent for most of the time (until the graphics card gets a hiccup and randomly goes to 100% fan speed for 2-3 seconds every 2 hours).
 
The intel website says the 265k uses more power than the 12700k so there goes that theory. I guess I'll just wait for reviews and hope for the best. If the newest gen cpus from amd and intel are both minimal upgrades this time around I guess I'll just limp along until the next series with my 12700k and 4090 lol.
I assume you're looking at the stated max turbo power of 190W and 250W. Intel's claim is more about overall efficiency improvements. At the same power, the 265k should do better. That it can go higher means it could extend its lead. We'll have to wait and see how this turns out in practice. I saw 9800X3D mentioned earlier, so presumably this is gaming focused? Edit: 12700k is 8+4 with HT, 265k is 8+12 without. I'd expect to see quite a big MT perf uplift from the 265k.

As long as the PC isn't really old, CPUs were never a good reason to upgrade the whole PC. It's only that marketing pushes the whole "you need it" and people often spend money on pointless upgrades. Each generation gives +5-15% IPC. If your CPU is 1-2 generations back, then there is barely any difference in daily usage or even gaming (I'm not saying about something like P4 to C2D jump).
I went Skylake-X to Zen 4. That was quite a jump in low thread performance. But I'm still wondering what Zen 5 will bring. My goal has been best gaming experience without breaking the bank (no xx90). If dropping in a 9800X3D can give say 10%, it could still be worth it. This is not just in fps, but load times saw a major improvement from the upgrade. I never timed it but it was very obvious.
 
Last edited:
I went Skylake-X to Zen 4. That was quite a jump in low thread performance. But I'm still wondering what Zen 5 will bring. My goal has been best gaming experience without breaking the bank (no xx90). If dropping in a 9800X3D can give say 10%, it could still be worth it. This is not just in fps, but load times saw a major improvement from the upgrade. I never timed it but it was very obvious.

Skylake-X to Zen 4 was a significant difference in access time, efficiency, and many other things. In some specific workloads, there is a huge difference.

The 9800X3D looks great in current leaks. However, I think I wait for the 9950X3D. I wanted to change the 7950X that I use for tests and reviews, but it's not worth getting the 9950X at the current price. It changes nothing except for a slightly higher performance. The wattage and other things are similar, so both are good for current tests and reviews. Once I get the 9950X, I won't need anything else for the next two years, maybe more.
 
MSI-LEAK-ARROW-LAKE-REFRESH.jpg
This just got posted elsewhere. Note this is for "K" parts. To my reading of it, Intel recommends systems default to "Performance" profile unless the mobo is too potato to support it, where it can fall back to "Baseline". The 9 tier parts also have a warranty busting "Extreme" setting.

I had to look up PL4, since we don't usually see that. Apparently it is a power limit that the CPU will aim to NEVER exceed. Presumably this is to avoid extreme short term spikes. PL1 is the long duration limit. PL2 is the short duration limit. Typically seconds to tens of seconds if different from PL1, but in enthusiast settings often set the same as PL1.
 
Last edited:
I don't know what to think so I'll just wait for reviews. If my noctua d15 isn't good enough to keep it under 80 in XTU memory test then I'm over it. Maybe I'll put the $ towards a better power supply and a 5090 instead.
 
And the D15 does that with a 12700K, eh? Impressed!!

Anyway, the temp value seems arbitrary and low, but, if that's how you roll, that's how you roll. :)
 
And the D15 does that with a 12700K, eh? Impressed!!

Anyway, the temp value seems arbitrary and low, but, if that's how you roll, that's how you roll. :)
It's not arbitrary. Without the contact frame it was occasionally throttling on warm days at 90c according to xtu. With fans tuned and a thermalright contact frame on an average temperature day it's around 78c give or take on full load in xtu mem test. I want 10c of wiggle room to account for summer temps and keep things healthy as long as possible. Having a tornado in my case is out of the question. The fractal torrent is very effective at air cooling with low noise so that helps a lot.
 
it was occasionally throttling on warm days at 90c according to xtu.
Hrm...I thought they don't start to throttle until 100C (shutoff at 110C). Your daily use is similar to that stress test load? Wowzas! Nothing I do comes close, but I know some loads can be.

I keep all my stuff on low too. I don't like things ramping up and getting loud... I feel ya on that front!! :)
 
Back