• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Arrow Lake Refresh

zandor said:
Wasn't much of an upgrade, and the i9s had fewer cores. 10th gen i9 had 10, 11th only had 8 like the i7. In other words the top end was worse in multi thread unless AVX512 was involved.

At any rate my point was Intel launched 2 architectures ~8 months apart, so I don't think a made by TSMC refresh is going to block a fabbed by Intel Nova Lake launch.
Click to expand...

TSMC didn't name Intel as a launch partner for 2nm, so either the initial batches are going to be from Intel's own fabs, or it's a lot more delayed than Intel is willing to publicly admit.
 
pippenainteasy said:
TSMC didn't name Intel as a launch partner for 2nm, so either the initial batches are going to be from Intel's own fabs, or it's a lot more delayed than Intel is willing to publicly admit.
Click to expand...
We were talking about Arrow Lake refresh on TSMC 3nm & Nova Lake on Intel 18A.
 
arestavo said:
Wasn't 11th gen worse than 10th gen? I remember looking at benchmarks and hard passing, and then buying a 10940X.
Click to expand...
Rocket Lake was very late to the party because Intel expected 10nm to be ready much sooner and didn't plan backporting Sunny Cove to 14nm early enough to be ready for 10th gen.
It had PCI-e 4.0 support and better single threaded performance.
It didn't directly translate to gaming performance showing some inefficiency in latency or cache subsystem and why having less cores and similar power consumption (and more single threaded power consumption) it wasn't exactly recommneded.
Also - and this isn't something many people are of - the failure rate of these CPUs is pretty high. Kinda Raptor Lake but without popularity it was swept under the rug.

Intel should have developed this architecture directly on 14nm and don't bet on 10nm so much. Then it would release sooner and be just better than it was.
Otherwise kinda pointless release. Alder Lake happened soon after suggesting that Intel could just as well release Sunny Cove at 10nm with more cores and higher clocks.

-------
LGA1851 - imho worst socket Intel released.
Single underperforming in games CPU series which didn't even use their own silicon.
Even though performance in games is worse than RPL I would have upgraded to Arrow Lake for excellent performance in programs if these CPUs were available for LGA1700.

Refreshing Arrow Lake makes no sense.
Instead of wasting money on such pointless releases they should focus on Nova Lake to not screw it up.
I fear Intel will come any day now and communicate how they are gonna screw Nova Lake... any day now...
 
arestavo said:
Wasn't 11th gen worse than 10th gen? I remember looking at benchmarks and hard passing, and then buying a 10940X.
Click to expand...

XoR_ said:
Rocket Lake was very late to the party because Intel expected 10nm to be ready much sooner and didn't plan backporting Sunny Cove to 14nm early enough to be ready for 10th gen.
It had PCI-e 4.0 support and better single threaded performance.
It didn't directly translate to gaming performance showing some inefficiency in latency or cache subsystem and why having less cores and similar power consumption (and more single threaded power consumption) it wasn't exactly recommneded.
Also - and this isn't something many people are of - the failure rate of these CPUs is pretty high. Kinda Raptor Lake but without popularity it was swept under the rug.

Intel should have developed this architecture directly on 14nm and don't bet on 10nm so much. Then it would release sooner and be just better than it was.
Otherwise kinda pointless release. Alder Lake happened soon after suggesting that Intel could just as well release Sunny Cove at 10nm with more cores and higher clocks.

-------
LGA1851 - imho worst socket Intel released.
Single underperforming in games CPU series which didn't even use their own silicon.
Even though performance in games is worse than RPL I would have upgraded to Arrow Lake for excellent performance in programs if these CPUs were available for LGA1700.

Refreshing Arrow Lake makes no sense.
Instead of wasting money on such pointless releases they should focus on Nova Lake to not screw it up.
I fear Intel will come any day now and communicate how they are gonna screw Nova Lake... any day now...
Click to expand...
Replying to both:

Rocket Lake's 11900K was 8 cores Vs. Comet Lake's 10900K with 10 cores.

The 10 core would beat the 8 core in multithreading loads, basically every time. The 8 core could only catch it or barely beat it, if you were overclocking it hard. And/or if it was using AVX512.

For other types of programs, Rocket Lake was generally better. And while sometimes it was a pretty big increase. Sometimes it was not.


For gaming, 11th Gen was initially a mixture of slightly better or regressions. IIRC, it was due to issues with core-to-core latency. Which they basically fixed over time, by tuning the microcode. And for games which enjoy better single core performance, it can show benefits.
However, the 10900k/10850K have a lot of cache. Which some games love, and that can keep 10th gen i9 winning.

Once you get down to i7 and i5: I would take 11th gen all day. As the core improvements express themselves more Vs. 10th gen i7 and i5, which did not have a big cache advantage, like the i9.

I personally had a 10700 non-K and swapped to an 11700 non-K, on the same motherboard.
 
Well, it's a little bit interesting to be honest. I'm curious how much faster it actually is head to head against a 14900K. Considering it uses a decent amount less power and that's the big improvement here. The only issue is the next gen is going to be better and this one requires a new CPU block and board. Let's see how it matures in a while. Also considering that since it's sales are going to be poor I'm guessing there will be some steep discounts on the platform so the value might there, maybe.
 
March 11th should be be interesting too bad there isn't a 290k. I guess reviews are at a later date. March 23rd or something
 
Last edited:
Hallock is insisting there are improvements which increase CPU limited gaming performance.
He also did a new video on Intel's APO software, in the past couple of weeks. Its strange to me how often they mention that app-----yet rarely update the list of supported games. I think there are only about 40, at this point in time.
Maybe they intend to develop APO more, with the launch of the refresh desktop CPUs and the new Nova Lake laptop chips.
 
$299 is great for the 270K plus! I was worried they would try and make it the new $400 chip like the 265K at launch.
 
Isn't the gaming performance mostly with P-Cores? There's no increase in Core frequency, only in the die-to-memory controller. Sucks to be the original Arrow Lake buyers who are stuck with the slower die-to-memory speeds.
 
Yea It's looking like the 270K plus is the new 14900K replacement except with better power efficiency if it truly is faster in gaming. All of a sudden this may perhaps get a bit interesting 🤔 lol
 
xDiVolatilX said:
Yea It's looking like the 270K plus is the new 14900K replacement except with better power efficiency if it truly is faster in gaming. All of a sudden this may perhaps get a bit interesting 🤔 lol
Click to expand...
This. Looks like my 265k is getting passed to the wife or kiddo. If this is better than 285k performance for 299 its a killer value.
 
I wonder if this chip can run Borderlands 4 I hear of mixed results on Steam page think its due to lack of Hyperthreading. Unless these people with a 285 K are all trying to run Borderlands 4 @ 4k.
Gamers Nexus reviews the dumbest games in their Benchmarks they benchmarked Starfield Steve Burke said the Game that nobody plays. Then why Benchmark it?
 
Comixbooks said:
I wonder if this chip can run Borderlands 4 I hear of mixed results on Steam page think its due to lack of Hyperthreading. Unless these people with a 285 K are all trying to run Borderlands 4 @ 4k.
Gamers Nexus reviews the dumbest games in their Benchmarks they benchmarked Starfield Steve Burke said the Game that nobody plays. Then why Benchmark it?
Click to expand...
Hardware Unboxed showed Arrow Lake with good performance in BL4

time stamped

View: https://youtu.be/d2hGLaQQpUk?si=3vcXhpP_5Ra0-G42&t=478

I think Techpowerup had it in their 9850X3D review, as well. But, it looks like they are currently re-doing how they present their gaming charts on their reviews. As the layout is different, and the chart loading is half broken.
 
workshop35 said:
This. Looks like my 265k is getting passed to the wife or kiddo. If this is better than 285k performance for 299 its a killer value.
Click to expand...
Yes it sounds like a dream come true and I wouldn't blame you for doing it especially if it's single threaded performance is faster than a 14900k but let me throw a curveball at you though how are we getting 64gbs of CU 8800mhz RAM? 🫪 haha
Jk you're probably already have your own ram in your current system but to squeeze the most out of somebody migrating from a older platform you probably want 8,000 or higher if you're going to buy a fresh set cuz if I'm not mistaken unlike AMD this new Intel platform scales a lot better with the faster memory if I recall correctly I haven't looked into benchmarks much because the platform was not exciting but now with this exciting news we're going to have to dig in a little bit now.
 
xDiVolatilX said:
Yes it sounds like a dream come true and I wouldn't blame you for doing it especially if it's single threaded performance is faster than a 14900k but let me throw a curveball at you though how are we getting 64gbs of CU 8800mhz RAM? 🫪 haha
Jk you're probably already have your own ram in your current system but to squeeze the most out of somebody migrating from a older platform you probably want 8,000 or higher if you're going to buy a fresh set cuz if I'm not mistaken unlike AMD this new Intel platform scales a lot better with the faster memory if I recall correctly I haven't looked into benchmarks much because the platform was not exciting but now with this exciting news we're going to have to dig in a little bit now.
Click to expand...
Arrowlake doesn't really benefit directly from the bandwith of the really high speed RAM. I think probably past DDR5 7200, the raw bandwidth doesn't matter for Arrow Lake performance.

The important part is the net benefit on latency. Particularly because Arrow Lake at stock settings, is very high latency.


*that said, its my opinion that the "AMDip", which some channels insist is real for certain gaming scenarios: is because AMD's memory controller is really low bandwidth.
 
chameleoneel said:
*that said, its my opinion that the "AMDip", which some channels insist is real for certain gaming scenarios: is because AMD's memory controller is really low bandwidth.
Click to expand...
I never looked too much into the "amdip" issue specifically but it definitely does bog down a bunch in some scenarios, especially loading new stuff like when you spin the camera around. It did this every time in FF14 it bogged down horribly which is why I never have switched my rig over to my AMD chips even though I have them. This kind of stuff is hard to put into a benchmark and test for as it's pretty specific and a lot of people that just play casually probably wouldn't notice.

A long time ago I remember somebody posted that as long as it can keep the cache filled X3D will dominate but if it's something where it can't then it is awful. If AMD could fix their horrible IO die then the perf would go way up.
 
xDiVolatilX said:
Yes it sounds like a dream come true and I wouldn't blame you for doing it especially if it's single threaded performance is faster than a 14900k but let me throw a curveball at you though how are we getting 64gbs of CU 8800mhz RAM? 🫪 haha
Jk you're probably already have your own ram in your current system but to squeeze the most out of somebody migrating from a older platform you probably want 8,000 or higher if you're going to buy a fresh set cuz if I'm not mistaken unlike AMD this new Intel platform scales a lot better with the faster memory if I recall correctly I haven't looked into benchmarks much because the platform was not exciting but now with this exciting news we're going to have to dig in a little bit now.
Click to expand...
48gb of 8400 before the crazy price spikes. Its been great so far, too bad its a dead platform after this.
 
chameleoneel said:
"reviews will land a couple days before these go on sale"

View: https://youtu.be/7QmVBH6CQ24?si=sJc34VMcqkATGKl3
Click to expand...

Hallock making some bold claims. The chart is showing decent gains compared to the previous chips but question is whether it's faster than Raptor lake. If it breaks even it's not going to be worth it cause CPU and board and water block is all gonna cost at least 600 not including ram and reusing previous ddr5 we already have. Yes it uses half the power potentially but it's a tough sell although the PC room will be noticeably cooler I still don't think it would push me over the edge to jump into it mainly because of the time and hassle and energy I would need to put into a completely new platform. It is nice no doubt just not sure how worth it it is to fit in all that work.

To be completely honest excitement for this actually just makes me even more excited for Nova lake.
 
Last edited:
I really wanna see if IBOT is marketing smoke or the real deal.

And then, what will Intel's long term commitment be to getting game profiles continuously added.
 
El_Capitan said:
Isn't the gaming performance mostly with P-Cores? There's no increase in Core frequency, only in the die-to-memory controller. Sucks to be the original Arrow Lake buyers who are stuck with the slower die-to-memory speeds.
Click to expand...

I'm pretty sure 200S Boost function already increases D2D from 2.1GHz to 3.2GHz on original Arrow Lake CPUs. So the default D2D speeds of the refresh have increased by 900 MHz, from 2.1GHz to 3.0GHz, which is still slower than what you already have with the non-refresh chips by just turning on 200S Boost. But it just works out of the box without having to turn it on like XMP. So for pre-builts which usually don't turn on XMP or other features out of the box, it will be slightly faster than stock OG Arrow Lake.
 
Last edited:
pippenainteasy said:
I'm pretty sure 200S Boost function already increases D2D from 2.1GHz to 3.2GHz on original Arrow Lake CPUs. So the default D2D speeds of the refresh have increased by 900 MHz, from 2.1GHz to 3.0GHz, which is still slower than what you already have with the non-refresh chips by just turning on 200S Boost. But it just works out of the box without having to turn it on like XMP. So for pre-builts which usually don't turn on XMP or other features out of the box, it will be slightly faster than stock OG Arrow Lake.
Click to expand...
Yep. 200S Boost mode bumps up both die to die and fabric clocks. The other possibly big thing is 200S Boost mode requires XMP to be on. I'm not sure what happens if your ram simply doesn't have XMP support.

All of this can be done with manual overclocking, so owners of original models aren't exactly screwed and can probably keep up unless Intel quietly made some changes to the physical design to improve performance beyond just adding a few e-cores on the u7 270k.

If they're not going to make a higher clocked replacement for the 285k they should make a 286k with the same clocks and the boosted D2D & fabric speeds. Oh that note, we're going to need a 386k and a 486k.
 
The 250k is really impressive for $200, I'll probably pick a 270k when they go on sale.
 
One big thing I noticed looking at reviews is that you really have to pay attention to whether the reviewers ran the original Arrow Lake chips bone stock or flipped on 200S boost mode. TechPowerUp enabled 200S boost mode, Tom's hardware apparently didn't. With boost mode on the 285k is really close to the 270k. Without it the 285k gets hammered unless manually OC'ed. So not replacing my 285k. I have 200S boost mode on, so it'll run very close to a 270k. If you're OCing manually I bet it's a silicon lottery crapshoot between a 285k and a 270k. No way in hell I'd buy a 285k now with the 270k priced at $300.

I think the real story with this update is the price/perf improvement. Price-wise they're up against the 9700X and 9600X and have a lot more punch in loads where the e-cores are useful.
 
FF14 was a huge problem for Arrowlake. No amount of tweaking would help it much. Even DDR5 8400 + 200S boost + 5.5ghz Pcores/4.9ecores for my 265K---- would only very slightly improve framerates.

So, IBOT appears to be the real deal. Hopefully Intel is aggressive in frequently adding games to IBOT. And also......hopefully they don't forget about Arrowlake Refresh after this year....

1774335618932.png
 
Considering the increase in power use on the 270K------I can see why they didn't release a 290K right now.

A. they get a little "win" in having pretty great overclocking for the 250K and 270K

B. A 290K is going to need to be binned with very efficient golden chips.
 
chameleoneel said:
FF14 was a huge problem for Arrowlake. No amount of tweaking would help it much. Even DDR5 8400 + 200S boost + 5.5ghz Pcores/4.9ecores for my 265K---- would only very slightly improve framerates.

So, IBOT appears to be the real deal. Hopefully Intel is aggressive in frequently adding games to IBOT. And also......hopefully they don't forget about Arrowlake Refresh after this year....

View attachment 793296
Click to expand...
AMD/X3D is kind of crap on FF14 in reality though, jerky/stuttery. That's why I've never used my 5800X3D or 7800X3D as my daily driver so they just kinda chill in the other room and why I never got the 9800X3D.
 
Last edited:
You must log in or register to reply here.
Back
Top