News Intel's next-gen Nova Lake CPUs will seemingly use a new LGA1954 socket

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
I'm sure the whole 10 people who bought into ArrowLake Desktop will be mildly upset? Surprised? Perhaps won't feel anything if they bought into ArrowLake?

Bad cynicism aside, I think Intel may be trying to differentiate itself from AMD's mainstream by, perhaps, increasing the platform (not CPU) richness instead. Which is good. Unfortunately, at the end of the day, what sells the platform is the CPU first and foremost.

I like ArrowLake's platform; it is very feature rich. But the CPU man... The CPU.

Regards,
 
I'm sure the whole 10 people who bought into ArrowLake Desktop will be mildly upset? Surprised? Perhaps won't feel anything if they bought into ArrowLake?
I certainly bought my 265K assuming it was one and done. Though that is why I ended up with a 265K instead of 245K with thoughts of upgrading.
I like ArrowLake's platform; it is very feature rich. But the CPU man... The CPU.
Outside of gaming (and AVX512) it's a better choice than AMD at every price point below the Ultra 9. That combined with the platform are why I ended up going with it even though I was underwhelmed by the performance (I still wonder what ARL looks like with better clock scaling).
 
Funny thing about that...
  • Intel didn't change sockets between Sandybridge vs. Ivy Bridge, yet Ivy Bridge increased PCIe support to 3.0.
  • Same thing for Ryzen 3000 adding PCIe 4.0 on AM4.
  • Same thing for Rocket Lake on LGA1700 (although, Comet Lake was originally meant to have it, before Intel pulled it back).

Oh, and about RAM? Skylake supported both DDR3L and DDR4. Alder Lake & Raptor Lake support both DDR4 and DDR5.

So, this shows a socket change isn't necessary, although the availability of some features will be board/chipset-specific.
I think what @closs.sebastien meant was not that the socket needs to be changed when such improvements occur but that the only time closs.sebastien upgrades a CPU is when one of those improvements comes along, which necessitates a motherboard change anyway because the old motherboard wouldn't have those new features.

I went from a Ryzen 1800X to a Ryzen 5700X3D. That's almost an ideal use case and to make it better I had 3200MHz memory that I had bought years before but was running at 2400MHz because of CPU support, so the 5700X3D was a huge upgrade for me. And it was cheap.

But I almost regret it. My 1800X still worked well; I don't usually notice a difference with the new CPU and my games are still GPU-limited. And I don't want a lot of the newer GPUs with PCIe 4.0 or 5.0 x8, because my motherboard supports PCIe 3.0 x16. (Yeah, I know, in most games I can get 97% of the performance.) I already had one of the fastest PCIe 3.0 SSDs. To get my computer where I was hoping I really need a whole new platform so the new CPU for the old platform wasn't a good value.

But I had a friend go from Ryzen 2600X to 3600X and he was very happy with the change. A particular game saw like a 100% performance boost for him.
Yeah, over a year ago I changed out my Ryzen 3200G for a 5600X. I did so because the 5600X was affordable enough and seemed on paper to be a drastic increase in performance. In practice I basically haven't noticed a difference at all. I don't game at high frame rates and am not often doing computation-heavy tasks. Maybe my code compiles faster? But it's not like my system ever felt slow on the 3200G. I'm not going to feel the need for a PC upgrade for a very long time, unless a cheap GPU that is more than twice as fast as my 1660 Super becomes available.
 
What are you basing that on? AM4's history? AMD's claim of AM5 support until 2027? Neither of those guarantee that any further generations will use AM5.
Fair point BUT AMD has a history of supporting sockets for far longer than Intel going a fair ways back, not just AM4 (had similar luck with socket 939, AM2/2+ even socket A was long lived via two gens/ refreshes of thunderbird/barton). Yes you might not get the newest gen stuff each time or it might have been limited to certain boards (which is true of AM4) but some support is better than none at all. And if rumors are true Zen 6 will be a drop in replacement as well for AM5 but that is yet to be confirmed officially, least anywhere I have looked. Point is AMD's past speaks volumes for its future as does Intel's. Name the last time Intel supported a socket for three generations? Exactly twice by my count (or gen, refresh, gen die shrink for haswell/devils canyon/broadwell or 12, 13, 14 series) but maybe you can remember something I don't recall, truly I hope I am wrong here.

The larger point is garnering the good will of enthusiasts/diy community is something Intel REALLY needs at the moment. And something as easy as going with three generations or two generations and a refresh can help a lot with Intel's mind share. I suspect IF AMD doesn't bring Zen 6 to AM5, which is something I find highly unlikely currently, but lets just say Zen 6 doesn't go AM5. A refresh of Zen 5 is still extremely likely at the least with claims of 2027 support.

Heck look at AM4 again. We got Zen, Zen+ (refresh), Zen 2, Zen 3 and Zen 3 refresh (XT models) on at least some motherboards. That is honestly pretty darned epic in my book for consumer support. And if AMD doesn't support AM5 with Zen 6, I promise I'll be just as eager to call them out on it. I have zero loyalty to these companies. What I am loyal to is bang for my buck, overall performance, loyalty to consumers and for sometime now...Intel just hasn't had those at the same time.
 
Last edited:
I went from a Ryzen 1800X to a Ryzen 5700X3D. That's almost an ideal use case and to make it better I had 3200MHz memory that I had bought years before but was running at 2400MHz because of CPU support, so the 5700X3D was a huge upgrade for me. And it was cheap.

But I almost regret it. My 1800X still worked well; I don't usually notice a difference with the new CPU and my games are still GPU-limited. And I don't want a lot of the newer GPUs with PCIe 4.0 or 5.0 x8, because my motherboard supports PCIe 3.0 x16. (Yeah, I know, in most games I can get 97% of the performance.) I already had one of the fastest PCIe 3.0 SSDs. To get my computer where I was hoping I really need a whole new platform so the new CPU for the old platform wasn't a good value.

But I had a friend go from Ryzen 2600X to 3600X and he was very happy with the change. A particular game saw like a 100% performance boost for him.
Are you sure your system is working properly? You should definitely be noticing a difference between a Zen 1 and Zen 3---especially one with 3D VCache vs one without. Only thing I can think of otherwise is that you play a lot of games that are brickwalled by the GPU.
 
hum, when we change the cpu, it is mainly to change the platform, like going from pcie3/4 to 5, from ddr3/4 to 5, to get new usb (c), etc..
so no matter the socket, the motherboard is changed any way.

do you know people who only change the cpu when they do a big upgrade? or change the cpu every year? I don't.
Everybody I know with an AM4 motherboard put more than one CPU in it.
 
Heh, I wonder if they'll make a version with no compute chiplets that just has the 4 LPE cores!
: D
Reminiscent of the Athlon 3000G, or that dual-core Alder Lake chip. It would probably be worse than expected if the LPE cores have very little or zero L3 cache. It won't be worth it to anybody unless they are fishing a brand new LGA1954 motherboard out of the trash. Or mugging someone in the Micro Center parking lot.

If a chiplet has 8+16 on it, maybe 6+4+(LPE) would be the bottom option, like the 6+4 Arrow Lake Core Ultra 5 225. That is only one SKU and the two above it have 6+8, which ought to be the new minimum instead. There's no reason to disable 75% of the E-cores, and there's more potential combinations in the stack with two core chiplets.

Here's my prediction. It's probably wrong:
Ultra 3 - 6+8(+4)
Ultra 5 - 8+12(+4)
Ultra 5 - 8+16(+4)
Ultra 7 - 12+16(+4)
Ultra 7 - 12+24(+4)
Ultra 9 - 16+24(+4)
Ultra 9 - 16+32(+4)

I don't need 52 cores for a laptop or even most desktops, I'd much rather have longer battery life or a fanless, silent workstation.
Most anything heavy goes up in the cloud anyway.
Crazy stuff.
So you don't buy a Core Ultra 9. You get the U3 or U5 equivalent instead.

Intel may be a good choice for your "fanless workstation" if they have superior integrated graphics (may not be true once Zen 6 arrives, we'll see).
 
Most anything heavy goes up in the cloud anyway.
I dabbled with AWS EC2, about 6 months ago. I found a 4-core Xeon instance was quite inexpensive, but costs scaled pretty linearly with the amount of additional cores you wanted. By the time it was more powerful than what I'd be willing to run at home, it was also expensive enough that I'd want to use it somewhat sparingly. Keep in mind that these server cores are running at much lower clock speeds than what you'd use, so you probably need at least 48 vCPUs to match a 32-thread desktop CPU.

Basically, if you need to do a lot of computation and you're paying out of pocket, it's cheapest to have your own machine. For commercial users, the costs look a bit different, since they have no depreciating assets on the books and save a bit of money on sysadmins (although there's a certain amountof management of the cloud VMs that still has to come out of somebody's time).
 
  • Like
Reactions: snemarch
When I get a new CPU I also want other i.e. more advanced new hardware to go with it. I see no reason to recycle the old stuff.
This is typically me. I usually wait long enough between upgrades that the new CPU needs a new type of memory and supports a faster PCIe. As a matter of fact, I went from FPM SIMMs -> DDR -> DDR3 -> DDR5, in the machines I used for Windows. If I do grab a Bartlett Lake CPU, it'll be my first time upgrading the CPU and not also the motherboard.
 
  • Like
Reactions: closs.sebastien
Thats sorta the N100/N150's of today. they arnt much dififerent. If you get the 6W TDP version
Yes, but Alder Lake-N has L3 cache, unlike the LPE cores to date. That somewhat makes up for its single memory channel.

I have a N97 and it's good for my purposes. Actually, not as low-power as I was hoping. With only raising the package power limits and no overcocking, I can get the machine to use over 60W at the wall. That involves the GPU, though. The entire machine idles at 7.5W, at the wall.
 
  • Like
Reactions: usertests
Fair point BUT AMD has a history of supporting sockets for far longer than Intel going a fair ways back, not just AM4 (had similar luck with socket 939, AM2/2+ even socket A was long lived via two gens/ refreshes of thunderbird/barton). Yes I might not get the newest gen stuff each time or it might have been limited to certain boards (which is true of AM4) but some support is better than none at all. And if rumors are true Zen 6 will be a drop in replacement as well for AM5 but that is yet to be confirmed officially, least anywhere I have looked. Point is AMD's past speaks volumes for its future as does Intel's. Name the last time Intel supported a socket for three generations? Exactly twice by my count (or gen, refresh, gen die shrink for haswell/devils canyon/broadwell or 12, 13, 14 series) but maybe you can remember something I don't recall, truly I hope I am wrong here.

The larger point is garnering the good will of enthusiasts/diy community is something Intel REALLY needs at the moment. And something as easy as going with three generations or two generations and a refresh or more can help a lot with. I suspect IF AMD doesn't bring Zen 6 to AM5 which something I find highly unlikely currently but lets just say Zen 6 doesn't go AM5. A refresh of Zen 5 is still extremely likely at the least with with claims of 2027 support.

Heck look at AM4 again. We got Zen, Zen+ (refresh), Zen 2, Zen 3 and Zen 3 refresh (XT models) on at least some motherboards. That is honestly pretty darned epic in my book for consumer support. And if AMD doesn't support AM5 with Zen 6, I promise I'll be just as eager to call them out on it. I have zero loyalty to these companies. What I am loyal to is bang for my buck, loyalty to consumers and for sometime now...Intel just hasn't had those.
We will see some form of Zen 6 on AM5. AMD isn't going to change sockets until the next memory type is ready so DDR6.

And agree with your post 100%
 
Here's my prediction. It's probably wrong:
Ultra 3 - 6+8(+4)
Ultra 5 - 8+12(+4)
Ultra 5 - 8+16(+4)
Ultra 7 - 12+16(+4)
Ultra 7 - 12+24(+4)
Ultra 9 - 16+24(+4)
Ultra 9 - 16+32(+4)
I don't think they're going to increase the base SKU count that much again (ARL has 5, RPL 6, and ADL 7). One of the things Lip-Bu Tan has talked about is streamlining and this would be the opposite. I'd bet on there being two SKUs using dual compute tiles before dropping down to single.

There's also laptop SKUs to take into account and the leak about two compute tiles also cited laptop only having one. While Intel is hardly bashful about weird laptop CPU naming I'm not sure they'd sell an Ultra 5 (8+16) in desktop and then call it Ultra 7 or 9 in laptop form.
 
  • Like
Reactions: usertests
Yeah typical intel I guess. Still for non-enthousiasts (basically everybody rounded up) a new socket doesn’t matter one bit.

Since I got stuck with a 6700k on a 1151 socket though, with the “next” socket also being 1151 just a little different so I’d have to get a new mobo to get a new cpu, I’ve decided no more intel for me unless they make something so awesome I can’t ignore it.

I’m on a 9800x3d now so I don’t see myself upgrading anything much the coming year aside from possibly adding an ssd or two.
 
hum, when we change the cpu, it is mainly to change the platform, like going from pcie3/4 to 5, from ddr3/4 to 5, to get new usb (c), etc..
so no matter the socket, the motherboard is changed any way.

do you know people who only change the cpu when they do a big upgrade? or change the cpu every year? I don't.
I do believe this is about intels short platform life cycle. what you say is true however intel does not align to the above philosophy, as traditionally their changes are much aligned with "they make money off of chipsets traditionally unlike AMD which for many years relied on 3rd parties now of course they also make their own chipsets" ... most of the functionality you mentioned now has moved to the integrated CPU controller where as years ago it was in south bridge controllers or north bridge controllers of your chipset. Finally, you can change the chpset and still keep socket compatibility as see comment most controller functionality resides in the CPU itself and giving people an upgrade path for several generations.

the point by Intel forcing mb changes/chipset changes it's leaving itself open to competition. they c
 
Intel needs to learn the weakness of its market position and stop doing this to its customers.

Sockets that don't last long is a major fail. I get that Intel has done this for a long time and it also takes time to change culture. That's fair. I hope this new 1954 will have twice or longer the lifetime of its predecessor.

Its fair to say that that no socket lasts forever. However Intel's socket strategy causes abandonment issues.
You need to learn the market, not Intel. Intel sells the overwhelming percentage of its CPU's to OEM's which end up in businesses. Of the 100's of desktops the company I work for have purchased, not a single one has been upgraded, nor has it even been considered. We keep the desktops 4 to 5 years and then send them to recycling and buy a new one. By replacing the socket every couple of years, no matter when the cycle comes up to replace a desktop, we're not going to be purchasing a platform that is more than a couple years old which matters if we're using it for 5 years. Not a single one of those desktops has been AMD based either, since there is no compelling reason to switch.

Intel caters to the business world, because that's where the volume/money is. OEM's want something new to sell every year. Upgradability is of zero importance to them. Lenovo shipped nearly 17 million PC's in Q4 of 2024, almost all of them Intel. You want Intel to "learn their weakness" and give a damn about the 3 people in this thread that upgraded their CPU? The DIY market is irrelevantly small.