• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

Arrow Lake Retested with Latest 24H2 Updates and 0x114 Microcode

Now that we have final BIOS and ME package v19.0.0.1854v2.2 (at least on Asus Motherboards), maybe one last retest @W1zzard ?
 
Why do it now at this stage? Just wait 4-5 more days and wait for the Intel's final powerpoint slide and then review to cross check against the actual claims at this juncture. No sense in rushing out a rushed review again.

Also the "Final" bios is only for Asus, Asrock is still on a "beta" release that allegedly spikes the latency as of Jan 2 (latest is Dec 26). :(
 
@W1zzard to be safe id recommended updating the bios to the latest and then update ME FW using MoKiChU ME FW Update from the ROG forum.

Alternatively you could use the motherboard download page to download the Asus ME FW Update tool, I'd recommend unzipping and from the FW folder manually flashing with FWUpdLcl64 -FORCERESET -ALLOWSV -F ME.BIN

I'm not sure if the Asus motherboard download page update tool uses -ALLOWSV by default, It probably does but it's not something I can test currently.

ALLOWSV does what it sounds like, allows flashing the same version, because the 19.0.0.1854v2.2 is still 19.0.0.1854, it should fail without using -ALLOWSV even if the sub-firmware is newer or different.
At least that has been my experience with prior ME versions and updates.

I expect there will be zero performance difference between earlier 19.0.0.1854 and the latest build.


Everyone else, make sure all your drivers are up to date including the Intel MEI drivers.
 
Last edited:
Regression in gaming is a tiny piece of the picture , gaming isn't be be all end all of computer use. The only thing wrong with arrowlake is pricing



I personally do hope HT variants make a return with a more efficient implementation because i literally see a 33% jump in performance for a tiny increase in thermals while doing my 3d render work on my 5700x , intels implementation hasn't changed much since they first introduced it , SMT from amd is actually a technically superior implementation of the concept.
Dude, I don't know what you are talking about, but I consider i7 and i9 (and their equivalent) of CPU'S explicitly made for f***ng gaming, or did I missed something in last 18 years? Especially, when the company who sell and design those products do tell so - also explicitly, or you mean Intel meant something else by "fastest gaming CPU on planet"? Well we can say it's a matter of interpretation - that's not the point, the point is - we have a problem if i (the consumer) buy something as such but - you (as vendor) do deliver a product that does not compute those communicated to, the consumer, very explicit claims of performance, and you take my money. Got it?
 
Last edited:
Didn't Intel's hallock say (2nd interview) they would update the 285k performance slides at CES so we have a clearer point of comparison now?

Its been 2 days, and nothing yet? From anyone including reviewers?

Was that just not done or is no one reporting on it?
 
You can't fix design flaw by software update, right?
 
You can't fix design flaw by software update, right?
if it happens once: its a design feature
if it happens twice: its design philosophy
if it happens constantly: its a software bug :laugh:
(similar to modern financial institutions)
 
So there's still light at the end of the tunnel, after all! For all you non-believers Intel presents January update for Arrow lake. Other article here.

1736409741423.jpeg


My remarks:
- more than 50% single core uplift in Cinebench 2024 is worth like 2-3 generational IPC improvements;
- 17.5% (on average) gaming FPS uplift is also too good to be true;
- 30% DDR memory latency reduction without physical (architectural) touch is close to miracle, but let's see;
- comparison baseline "Jan.'25 Config" vs. "Reviewer config" can basically mean anything :laugh:, pick the worst Arrow Lake review and miracles will be easily explained;
- based on experience with Intel's marketing materials, take this Intel statement with a 25t wagon of salt.
 
So there's still light at the end of the tunnel, after all! For all you non-believers Intel presents January update for Arrow lake. Other article here.
(snip)
I will wait for the next architecture after Arrow Lake before I take another serious look at Intel, I have little confidence that Intel can solve its long-term problems anytime soon. AMD offers less socket changes, less confusing product layouts and naming, and AMD products are competitive with Intel when not outright crushing Intel chips.

Personally I doubt Intel will realize the promises that are made on paper with Arrow Lake.
 
Personally I doubt Intel will realize the promises that are made on paper with Arrow Lake.
Arrow Lake is not the only instance of situation when Intel made promises and those were not even (half)-achieved.
Unfortunately, only a very few tech news sites roast them for that. Maybe people like being lied to.
 
What I don't understand about Intel is how they can price these more than AM5 chips like the 7950X much less other choices for even less money. The MB support on AM5 is also a huge advantage. Intel has done enough in the trust space that only the most ardent users still recommend Intel. If you bought it though you are not wrong. Enjoy your purchase.
 
What I don't understand about Intel is how they can price these more than AM5 chips like the 7950X much less other choices for even less money.
The i9 is as fast or faster and more power efficient in every single use case scenario, from ST to MT to gaming or mixed. That's according to the original TPU review. So why would they not charge more than the 7950x? The i9 is a much better product.

The only thing that will be better is the 9950x3d which is also probably going to be more expensive.
 
The i9 is as fast or faster and more power efficient in every single use case scenario, from ST to MT to gaming or mixed. That's according to the original TPU review. So why would they not charge more than the 7950x? The i9 is a much better product.

The only thing that will be better is the 9950x3d which is also probably going to be more expensive.
Arrow Lake is more power efficient than the 7950X?
 
Arrow Lake is more power efficient than the 7950X?
According to TPU (and other reviews but I focused on this one) yes, on pretty much everything. Multi single gaming light loads.
 
The i9 isa much better product.
I'd do wording a bit more carefully. Arrow Lake still has some problems with Win11 that cause BSOD or similar level failures and some problems do still persist even after Intel stated how they fixed 7 or 8 problems and all should be okay now. You can clearly see how Arrow Lake was unpolished at launch - they fixed how many serious bugs? Still, Arrow Lake has been there just for 3 months, who knows what will be found out in months to come. Zen 4 is more mature. Better to have a bit slower but stable CPU than a bit faster and problematic one.
 
I'd do wording a bit more carefully. Arrow Lake still has some problems with Win11 that cause BSOD or similar level failures and some problems do still persist even after Intel stated how they fixed 7 or 8 problems and all should be okay now. You can clearly see how Arrow Lake was unpolished at launch - they fixed how many serious bugs? Still, Arrow Lake has been there just for 3 months, who knows what will be found out in months to come. Zen 4 is more mature. Better to have a bit slower but stable CPU than a bit faster and problematic one.
Im just going off TPU's review, I haven't had any personal experience. According to TPU's review, its better than the 7950x at everything while consuming less power. It's clearly the better product. The 9950x is a lot more competitive and it's a wash between the 2.
 
Im just going off TPU's review, I haven't had any personal experience. According to TPU's review, its better than the 7950x at everything while consuming less power. It's clearly the better product. The 9950x is a lot more competitive and it's a wash between the 2.
You confuse me sometimes. Littlebro gave you a plausible reason and then you admit that you have no personal experience but come back with the exact same argument?
 
You confuse me sometimes. Littlebro gave you a plausible reason and then you admit that you have no personal experience but come back with the exact same argument?
You questioned whether it was more efficient than the 7950x. I think we solved that, right? So what are you confused about
 
I mean I can undervolt/tune both, not really fair to suggest tuning one when the same refinements could be perhaps had on the 285K too? I'm not really asking for that though, looking for more of a general out of the box (with updates, obviously!) overall experience - I'm not looking to spend hours and hours testing stability here to accomplish this, nor am I looking to run a mini heater (my 13900k right now is running while respecting Intel Limits, i.e. not to exceed 253watts).

The 285k also beats the 9950x in some workflows - and is neck and neck in others - it's just unclear what the power draw is during those specific workflows. Maybe even over the duration of the test? Dont think anyone measures that though - useful metric for me, but not to the benchmarkers it seems.

this video seems to literally be the only video or benchmark that directly compares the results while showing power draw in that benchmark. But obviously it's a month old video so none of the fixes are taken into account here... so I imagine it can only get better here? At least as far as gaming is concerned...

Thoughts on the above video? Anything I'm overlooking here? Even if I look at this video in isolation vs the non gaming performance benchmarks of 285k vs 9950x, 285k wins it.... right?


Thanks for the lengthy explanation and the picture.
And yeah I'm planning on waiting for the field 2of2 - but coming from Asus board to this Asrock z890 Aqua - their Bios pages don't actually show the ME version though in each bios. It just states: "Update Intel ME version." with no details. Is this somewhere that I'm not seeing? I even downloaded that bios right now and opened the zip file, theres no changelog or .txt file to peek in lol

I'm just a bit confused as I'm not sure if you watched the full part2 of the video, but the Intel guy acknowledged that 0x114 is out, but its useless without the secret sauce which, as we know is coming in January and is being tested by their partners right now. So are you suggesting that it's possible that the Asrock bios currently out right now, marked as 0x114, with no mention of "v2.2" in the ME firmware, was actually based off the 2.2 toolkit? So everything may already be as good as it gets possibly? I know the proper answer is to just wait at this juncture, just wondering about the possibility. Thanks again!

The original review has exactly what you're looking for.

1736644088000.png
1736644107079.png
 
What I don't understand about Intel is how they can price these more than AM5 chips like the 7950X much less other choices for even less money. The MB support on AM5 is also a huge advantage. Intel has done enough in the trust space that only the most ardent users still recommend Intel. If you bought it though you are not wrong. Enjoy your purchase.

I think it depends on the market, I'm in Italy so not in the giant US/Canada/Germany/UK market and here on Amazon.it a 7950X is priced at 520€, the 265K at 455€, the 7700X and the 245KF both at 299€. Looking at these charts:


the 7700X seems not a clear gaming/application/benchmark winner over the 245K. The 7600 with its 80-100€ less seems a better deal.

It's also a motherboard price point, for eg. I'm in the market for an ITX motherboard and a CPU, and the AM5 ITX on Amazon.it seems to start at 270€ with the MSI B650. The ASRock A620 Lightning isn't sold directly from Amazon. We have to see what will be the prices of B860 motherboards I think next week.
 
Last edited:
I think it depends on the market, I'm in Italy so not in the giant US/Canada/Germany/UK market and here on Amazon.it a 7950X is priced at 520€, the 265K at 455€, the 7700X and the 245KF both at 299€. Looking at these charts:


the 7700X seems not a clear gaming/application/benchmark winner over the 245K. The 7600 with its 80-100€ less seems a better deal.

It's also a motherboard price point, for eg. I'm in the market for an ITX motherboard and a CPU, and the AM5 ITX on Amazon.it seems to start at 270€ with the MSI B650. The ASRock A620 Lightning isn't sold directly from Amazon. We have to see what will be the prices of B860 motherboards I think next week.

I understand what you are saying. I live In Canada we have stupid prices for PC parts thanks in part to our distributor network.

This is the price of the 245K where I live


Meanwhile this is the 7600


It makes no sense that the gulf is that large. This is partly why the sales of these CPUs are so low. It reminds me of the MSI Claw.


You should look at Gigabyte. I was able to get a A620! for like $140 Canadian. That is probably like 110 Euros. Too bad you don't have Newegg. Another place you can check is Groovesland. They ship free Worldwide and might have some good prices for MBs.

You are right about the 7700 it is the same $439 as the 265K but the 7900 is only $40 more for 18 more threads. There are so many choices. You could save about $100 and go AM4 as the 5900X and 5950X are both well priced.
 
Last edited:
You should look at Gigabyte. I was able to get a A620! for like $140 Canadian. That is probably like 110 Euros. Too bad you don't have Newegg. Another place you can check is Groovesland. They ship free Worldwide and might have some good prices for MBs.

You are right about the 7700 it is the same $439 as the 265K but the 7900 is only $40 more for 18 more threads. There are so many choices. You could save about $100 and go AM4 as the 5900X and 5950X are both well priced.

First of all thank you for your time and reply. Another problem here is the availability since Amazon.it is the only store with a real and reliable warranty.

The Gigabyte A620I is not sold directly by Amazon right now, but the main "problem" of that board is its only 1x M.2 Slot, honestly a deal breaker for me since in a ITX case the space is gold and would like a second slot. Same for the Gigabyte B650I. The next step for an AM5 motherboards is the MSI B650I at 264€ or the ASUS B850-I at 350€.

Meanwhile Amazon just opened the pre-orders of the ASUS Intel B860-I motherboard at 299€, that would indicate that Intel is cheaper here.
 
Last edited:
285K is indeed newer product than 7950X and has much more cores. That being said, it is destined to win in multi-threaded workloads.
285K is on 3nm node while Zen 4 is 5nm node, so 285K must be more efficient. What is bad is by how much only is 285K more efficient given the huge node progress ...
285K should have been beating even Zen 5 in terms of efficiency but this does not happen.
EDIT: 285K beats 9950X in terms of efficiency on idle, in gaming and in single-threaded workloads, but not in multi-threaded workloads. (Based on data in TPU's review of 285K.)

Thus, I would not say Arrow Lake i9 is much better product. It is superior to 7950X, that's true, but should have had even better efficiency and more performance given the 285K's node advantage.
When it comes to support, Arrow Lake is definitely not better product, it was a disaster at the time of release and as I mentioned above, there are still problems that persist.
While some people are still reporting errors, one dude reported that latency of 285K with 0x114 is much worse.
Unfortunately, we are unable to see whether this latency is really worse, because it was not tested here along with synthetic memory benchmarks (AIDA64).

Anyway, you and also many others here keep missing one quite serious point:
Arrow Lake was retested with newest microcode(s) while there was no such retest done for AMD's Zen 4 and Zen 5.
AMD, too, released updates to improve Zen 4 and 5 performance in Windows 11 (improved pre-fetching and scheduling).
Hypotetically, we might observe comparable or even better 7950X's performance with newest microcode and OS updates but this was not tested.
 
Last edited:
285K is indeed newer product than 7950X and has much more cores. That being said, it is destined to win in multi-threaded workloads.
285K is on 3nm node while Zen 4 is 5nm node, so 285K must be more efficient. What is bad is by how much only is 285K more efficient given the huge node progress ...
285K should have been beateing even Zen 5 in terms of efficiency but this does not happen.
There is no such thing as it "should", but fact of the matter, it does beat Zen 5 as well in efficiency. I know data ain't really important when it comes to bashing Intel, but ill share the data just in case.

1) Idle power draw - the 9950x draws 38% more power
2) ST efficiency - the 285k is 36% more efficient. That's absolutely insane btw.
3) MT efficiency - they are on par, 0.0008% difference between them.
4) Gaming efficiency - the 285k is 12% more efficient.

All the data are taken from this review

 
Back
Top