• 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.
]-[itman said:
That's Intel's spin, it isn't really logical though. It's a cost cutting maneuver. Intel is desperately trying to cut costs as much as possible, this is one of those cuts.
Click to expand...
18A is quarters away from HVM and is the major node Intel has been pushing for wider use. Too late to use for Arrow Lake but could be in line for its successor. If money was unlimited then going ahead with 20A could still be a de-risk move. So yes, it is a money saving move (Tom's say half a billion), but one that does make sense.

infinitevalence said:
I can't decide if Intel is done and will never retain it's position or a hell of a good long-term stock buy.
Click to expand...
It could still go either way. The fuel warning light is on but the next gas station is on the limit of remaining range. Can they make it? Tune in next week to find out!
 
mackerel said:
It could still go either way. The fuel warning light is on but the next gas station is on the limit of remaining range. Can they make it? Tune in next week to find out!
Click to expand...
Every car goes past "0 miles remaining"... :p
 
I would like to think that they will learn a lesson from this about stagnating development and focusing on investors is bad for the long term. But I dont know that they can realize that.
 
They never stopped development. It's a common misconception since many only look at desktop where new product may be seen as lacking, but there's a big difference between not trying at all, and trying but failing to deliver. I don't feel their investor focus is out of whack with the rest of the industry. AMD and Nvidia don't feel much different in that respect. Intel's current main problems were kicked off from around 10 years ago. For a large part, if it could go wrong, it did. They are approaching the end of that recovery process and isn't something that can be fixed overnight.
 
mackerel said:
They never stopped development. It's a common misconception since many only look at desktop where new product may be seen as lacking, but there's a big difference between not trying at all, and trying but failing to deliver. I don't feel their investor focus is out of whack with the rest of the industry. AMD and Nvidia don't feel much different in that respect. Intel's current main problems were kicked off from around 10 years ago. For a large part, if it could go wrong, it did. They are approaching the end of that recovery process and isn't something that can be fixed overnight.
Click to expand...

They definitely took their foot off the Gas while AMD was flailing around with Piledriver. But they bet heavy on 10nm and under delivered, they are still selling 14nm parts because 10nm has bad yields.

My investor focus was specifically all the stock buy backs, and the over hiring in Covid because line goes up, line always go up. I personally dont see ARC as a bad bet, just bad timing, I think there is still room to return ARC to the market and make it not suck. Firing the Seattle CPU team was probably a big mistake and will end up creating a viable RISC V general CPU, it also puts all their eggs in the Isralie design team for any future CPU advancements.

I also worry that relying on TSMC long term for cutting edge fab will reduce their competitiveness which was in part due to the fab capacity they used to have in the 14nm era.
 
infinitevalence said:
they are still selling 14nm parts because 10nm has bad yields.
Click to expand...
Such as? Desktop since 12th gen is on 7. Mobile moved off 14 since 10th.

Pat has previously said they expect to be "peak TSMC" around now and as 18A comes online will bring home more leading edge stuff.

If RISC V will ever take over high performance CPU space is up for debate and so far off it is irrelevant at this point. Look at all the similar claims put on Arm which has not managed to kill off x86. Apple is about the only arguable success story there, and I'd attribute that more on Apple than Arm.
 
If you had some old stock lying around doesn't sound like an Intel problem. About the only exception I can reasonably think of is if it was a long term support product which doesn't need to be cutting edge (e.g. embedded or industrial), or some non-performance part that doesn't need to be leading edge. Official retirement of products can be a long longer than people expect, but volume usually has died off long before that point.
 
I have no interest in digging further into your business practices, but it sounds like an edge case that you chose to buy older gen from Dell. Not representative of the general/wider market. The mainstream desktop offering has been Intel 7 for 3 gens. Mobile has 10/7 offerings for 5 gens. 14nm is a distant memory.
 
265K looks like it's going to be a nice upgrade for me. 8 extra E-cores with higher ipc for the same power usage as a 12700k. Also gonna be going from ddr4 to ddr5, whatever the fastest is that the chip/mobo can handle with stability so I won't be gimped any more by bandwidth. Nvme 5.0 should be nice too but whether 15K Mb/s read speed feels faster in the real world remains to be seen.
 
Brando said:
265K looks like it's going to be a nice upgrade for me. 8 extra E-cores with higher ipc for the same power usage as a 12700k. Also gonna be going from ddr4 to ddr5, whatever the fastest is that the chip/mobo can handle with stability so I won't be gimped any more by bandwidth. Nvme 5.0 should be nice too but whether 15K Mb/s read speed feels faster in the real world remains to be seen.
Click to expand...

E-cores are generally used to support background tasks. If you don't use 20 web browser tabs or don't run 10 apps at once, then I doubt you see any improvement from more E-cores. What you may see is the improvement in the P-core IPC.

There is barely any improvement in user experience from PCIe 5.0 SSDs compared to PCIe 4.0. Out of some synthetic benchmarks or moving huge amounts of data between SSDs, there is barely any difference compared to the higher PCIe 4.0 series. Compare something like the Crucial T500 and T700 in various reviews, and you will see what I mean. These SSDs are mainly different in sequential bandwidth, while random bandwidth, access time, and even IOPS are not much worse on the T500. In benchmarks that simulate daily tasks like PCMark 10, both get almost the same scores.
Depending on what I have left from reviews and what I need in other rigs, I move between various PCIe 4.0 (often DRAM-less) SSDs and the top PCIe 5.0 series in my daily/gaming PC, and I see no difference. I keep PCIe 5.0 SSDs mainly because they are not worth selling (it's hard to get more than 60% store price), but I wouldn't buy any new PCIe 5.0 SSD because prices are still too high.
 
I'm pretty used to how this drive feels at 4.0 speeds so I guess we shall see.
Woomack said:
E-cores are generally used to support background tasks. If you don't use 20 web browser tabs or don't run 10 apps at once, then I doubt you see any improvement from more E-cores. What you may see is the improvement in the P-core IPC.

There is barely any improvement in user experience from PCIe 5.0 SSDs compared to PCIe 4.0. Out of some synthetic benchmarks or moving huge amounts of data between SSDs, there is barely any difference compared to the higher PCIe 4.0 series. Compare something like the Crucial T500 and T700 in various reviews, and you will see what I mean. These SSDs are mainly different in sequential bandwidth, while random bandwidth, access time, and even IOPS are not much worse on the T500. In benchmarks that simulate daily tasks like PCMark 10, both get almost the same scores.
Depending on what I have left from reviews and what I need in other rigs, I move between various PCIe 4.0 (often DRAM-less) SSDs and the top PCIe 5.0 series in my daily/gaming PC, and I see no difference. I keep PCIe 5.0 SSDs mainly because they are not worth selling (it's hard to get more than 60% store price), but I wouldn't buy any new PCIe 5.0 SSD because prices are still too high.
Click to expand...
At this point I may as well go for it. I have a bad ram slot on my existing motherboard and msi won't do advanced shipping for warranty so I'm just going to limp along with 1 stick of ram for a month or so and upgrade then sell the replacement.
 
Date leaks. We're getting close.

Pre-briefing for press on 7th. Maybe we get some early leaks?
Public announcement on 10th. (Thursday)
Sales and review embargo on 24th. (Thursday)

If rumours are right, we may be hearing of a 9800X3D some time soon too. Seems plausible if AMD want to retain their claim to gaming crown and try to divert attention away from Intel.
 
9000X3D chips are supposed to be at the beginning of next year. I assume they will be announced at CES, so the first weeks of January. At least it was mentioned some time ago. There are some more leaks around like both CCD will have the 3D v-cache, but no one leaked proper results, so I guess it's still some time for the release.

Regarding Z890 motherboards, I know most samples will be ready for shipping around mid-October. Some reviewers (maybe a few top websites) may get something earlier, but not most of them. I guess that only the so-called reviewers' packs (CPU/CPUs + one mobo and sometimes RAM) will be available earlier, and those go to the mentioned few top websites.
 
Woomack said:
9000X3D chips are supposed to be at the beginning of next year. I assume they will be announced at CES, so the first weeks of January. At least it was mentioned some time ago. There are some more leaks around like both CCD will have the 3D v-cache, but no one leaked proper results, so I guess it's still some time for the release.

Regarding Z890 motherboards, I know most samples will be ready for shipping around mid-October. Some reviewers (maybe a few top websites) may get something earlier, but not most of them. I guess that only the so-called reviewers' packs (CPU/CPUs + one mobo and sometimes RAM) will be available earlier, and those go to the mentioned few top websites.
Click to expand...
The last rumor I heard was the 9800x3d being announced this month and the 9900/9950x3d being after the new year. But time will tell. To not derail this thread too far, my understanding is the 9xxx chips are not selling and they want to push something forward that may show a difference and can potentially launch near arrow lake.
 
We are dealing with leaks, rumours and speculation. Nothing is certain. Like link above, there is a rumour that AMD will announce only 9800X3D soon. What else is happening soon? Arrow Lake. I don't think it a stretch the two will be related.

7800X3D has been the sweet spot for performance gaming focused systems since its release. Intel will likely compare that to ARL for gaming, assuming ARL does well in that area. I think AMD would like to give Intel a headache if they can say, yeah, but we got a faster CPU too. They don't have to ship it at the same time to mess with Intel.
 
Last edited:
You must log in or register to reply here.

Similar threads

mackerel
Relative IPC and efficiency of some Intel CPUs in Cinebench R15
Replies
3
Views
754
mackerel
mackerel
mackerel
Arm, NPUs, and Lunar Lake
Replies
2
Views
298
mackerel
mackerel
mackerel
Upcoming CPUs (2H 2024+)
Replies
2
Views
375
mackerel
mackerel
Super Nade
Athlon64 and General CPU Technology mini-refrence
2
Replies
30
Views
16K
jonspd
jonspd
Back