I do like that they are dialing back the TDP on these chips some. The 7000 desktop chips seems to push that a little too hard to be the fastest. A little too much of the Intel overclocked to the edge from the factory there. Nice to see them backing down from that so we don't start seeing issues similar to the i9's.
What would have REALLY been impressive is if AMD had also designed, announced, and released an AM5p Motherboard Platform. Something that is similar to the Desktop AM5 Motherboards but this time it is for Laptops, with the key feature being accessibility. So that can upgrade the CPU, RAM, SSD as the months go on. And maybe even a socket for dGPU.
I am aware of the Frameworks Laptop, but it would have been much better having this come directly from AMD. It would have really separated AMD as a premium option for laptops, against a walled/locked Intel platform, or the competitive ARM options coming in the future.
Just like how it was a massive upgrade to go from the first 8-core (r7-1700) to the last 8-core (r9-5800x3D) option on the AM4 Platform. It would have been great to see that on their AM4 Laptops (eg 4-core to 4-core for efficiency). But I understand that. Does not mean it cannot happen for AMD Laptops with this AM5 Platform though.
So yeah, AM5p Motherboards.... let's do that please!
I'd love to get those components user upgradable but soldered components reduces overall hardware production costs.
you also can't get unsoldered lpddr, which is also faster than dimm ddr.
as long as the ssd is removable and 2280 slot spec, I'm fine with it.
i just wish all laptop manufacturers give option not to include windows. i saw in hp website that it reduces the price by at least 100 dollar for the elite book model
Indeed. This is the trajectory of the market: CAMM. All of the benefits of lower power and higher frequencies of LPDDR memories with the added benefit of user upgradability. Link below:
I'd love it, unfortunately the world is moving more to throw away soldered designs, because screw reparability, upgradeability, or what some consumers want.
Intel screwed up does not mean AMD's Arch also follow same. AMD's Zen 3 processors ran at 1.4V, Intel Sandy to SKL all ran at 1.4V. None of them exhibit the damn instability or the degradation to the point of no return.
AMD needs to step up their TDP and wattage, the 9950X barely beats a 7950X. It's mediocre 41K CBR23, vs Zen 4's 38K. Compare to the Zen 3 vs Zen 4, the jump was over 14000points in CBR23.
AM5 socket is capped to 230W, no amount of PBO will exceed that. And PBO added Zen 4 goes to 40K and Zen 5 with Curve Shaper and PBO2 it will go upto 45K max.
AMD could have added more power, 50W for the R9, 25W for the R7 but they instead reduced it massively and also cut down the Base Clocks by a lot of MHz. Wonder if it's due to Intel's ARL dropping SMT / Hyperthreading.
Thanks for the article, Gavin. I am somewhat impressed by Zen 5, particularly what's going on in its floating-point domain and the hints of its encoding performance. Intel made an excellent move disabling AVX-512.
Intel will have the advantage starting with lunar lake in performance per watt. There is no getting around 3nm vs 4nm advantage. Also AMD famously does paper launches, they will launch it only to not find any products to buy. you still see there are a lot of laptops with AMD 7nm still being sold. It's their way of clearing the inventory.
Intel's problem is that they don't yet have AVX-512 on their E cores which makes things awkward. They theoretically could allow you to run it on the P cores but not the E cores but that has a lot of draw backs. You'd have to manage threads so they stayed on the P cores or they'd crash if they migrated to the E cores. You also might not get much benefit from it if you are limited to only P cores because you'd be loosing out on a lot of cores and the stuff that scales well on AVX-512 also tends to be the same type of things that scale well with more cores. Trying to run mixed code with AVX-512 on the P cores and run AVX-256 or something else on the E cores but that seems even more complicated. Intel does have a solution in the pipeline with AVX10 which will be available on the E cores eventually. The E cores will run it as combined 256 bit vectors so it will be slower but it will be able to run the same code.
The obvious solution was for them to implement AVX-512 on the E-cores, combining FPUs to do the job, and there wouldn't have been any problem. Apparently, AVX10 has issues of its own but I'm not too clued up on that at present.
That fixes the thread migration problem for single threaded workloads, but is only of limited help with multithreaded ones being a repeat of the old AMD shared FPU design.
I hear what you're saying: run away from Bulldozer's approach. I meant it in the sense of combining the four 128-bit vector paths in a single Skymont core to make up 512 bits, similar to Zen 4's manner of joining two 256-bit paths. Granted, Crestmont had only three 128-bit paths, making it problematic.
It is funny, especially considering that fast AVX-512 will become one of the hallmarks of Zen 5. These instructions are being increasingly used in everyday applications. For example, the SVT-AV1 encoder and dav1d decoder. This perhaps explains why Handbrake's performance was so high.
Good and interesting, but what about the i9-14900KS test redo with Intel Default settings? You told us 2 months ago that you'd redo them, and I think the recent news about possible issues with this CPU make it even more relevant :
"Gavin Bonshor - Friday, May 10, 2024 - link Don't worry; I will be testing Intel Default settings, too. I'm testing over the weekend and adding them in."
I'm saying this because it seems to me that, for the most part, CPUs that got the patches applied received them rather late in their lifetime, having run under possibly "wrong" settigns for quite a while. So the mitigations in such a case may have come too late to prevent damage from happening.
I'm wondering if there are statistics about that so far. That would be interesting to see if applying such mitigations changed anything on the "lifetime without failure" of the affected CPU, when applied early on its lifetime.
I don't doubt that the benchmark figures here are true, but I also believe that sites that need marketing funds can't be aggressive in pointing out manufacturers' problems.
I feel like AI9HX chip will be massive win. I noticed asus put them in vivobooks, with decent otherwise specs, no extra trash gpu, 90W. It feels like the best thing you can get right now, especially thin and light, as "paired" with 4060 is a joke. integrated can run like lowest tdp 4060 mobile. who in his right mind would want to double the power draw, for no other reasons than NVIDIA sticker and dedicated VRAM. If I find any laptop with 2 or 3 slots for ssd, any format, I will take that, if not, I guess that VivoBook it is. btw, It's really sad that small form laptops cannot find any space for a second m.2 slot. just why. Especially with all those PCIE empty with no gpu....
I would have liked to see an interim RDNA 3.5 for discrete cards too... Maybe a Navi 30 that would have powered the never released Radeon RX 7950 XTX. Might have been a nice stop gap. Maybe it was just too optimized for lower energy and mobile.
RX 8000 built on RDNA 4 looks to target low/middle of the market around Q1/Q2 2025 with no high end part (assuming Navi 48 is low, Navi 44 is middle).
What if it's just two sizes of chiplets, I haven't seen that AMD has said they are not targeting high end, we know they have been working on chiplet GPUs and now Nvidia is as well.
I'm on a i5-4570S for my main PC Nothing wrong with it really, except video deconding. Been hankering after an upgrade for a few years: Zen4, then Zen4-G... now it seems I've got to wait for Zen5 even probably Zen5-G but at some point I'll have to pull the trigger before MS takes Windows away from me ^^ My one concern is USB reliability. I use USB a lot, and last time I dropped AMD because of USB issues. That was 11 yrs ago, but I've read in some places that USB still isn't rock-steady the way it is w/ Intel.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
43 Comments
Back to Article
kpb321 - Monday, July 15, 2024 - link
I do like that they are dialing back the TDP on these chips some. The 7000 desktop chips seems to push that a little too hard to be the fastest. A little too much of the Intel overclocked to the edge from the factory there. Nice to see them backing down from that so we don't start seeing issues similar to the i9's.Gavin Bonshor - Monday, July 15, 2024 - link
AMD doesn't need to push higher wattages, which makes them more attractive from a perf/watt standpoint.Kangal - Tuesday, July 16, 2024 - link
What would have REALLY been impressive is if AMD had also designed, announced, and released an AM5p Motherboard Platform. Something that is similar to the Desktop AM5 Motherboards but this time it is for Laptops, with the key feature being accessibility. So that can upgrade the CPU, RAM, SSD as the months go on. And maybe even a socket for dGPU.I am aware of the Frameworks Laptop, but it would have been much better having this come directly from AMD. It would have really separated AMD as a premium option for laptops, against a walled/locked Intel platform, or the competitive ARM options coming in the future.
Just like how it was a massive upgrade to go from the first 8-core (r7-1700) to the last 8-core (r9-5800x3D) option on the AM4 Platform. It would have been great to see that on their AM4 Laptops (eg 4-core to 4-core for efficiency). But I understand that. Does not mean it cannot happen for AMD Laptops with this AM5 Platform though.
So yeah, AM5p Motherboards.... let's do that please!
zamroni - Tuesday, July 16, 2024 - link
I'd love to get those components user upgradable but soldered components reduces overall hardware production costs.you also can't get unsoldered lpddr, which is also faster than dimm ddr.
as long as the ssd is removable and 2280 slot spec, I'm fine with it.
i just wish all laptop manufacturers give option not to include windows.
i saw in hp website that it reduces the price by at least 100 dollar for the elite book model
TheinsanegamerN - Tuesday, July 16, 2024 - link
They could get around that with CAMM, which is modular but offers LPDDR speeds.Hifihedgehog - Tuesday, July 16, 2024 - link
Indeed. This is the trajectory of the market: CAMM. All of the benefits of lower power and higher frequencies of LPDDR memories with the added benefit of user upgradability. Link below:https://www.techpowerup.com/322754/lpddr6-lpcamm2-...
TheinsanegamerN - Tuesday, July 16, 2024 - link
I'd love it, unfortunately the world is moving more to throw away soldered designs, because screw reparability, upgradeability, or what some consumers want.Hifihedgehog - Tuesday, July 16, 2024 - link
https://www.techpowerup.com/322754/lpddr6-lpcamm2-...Terry_Craig - Tuesday, July 16, 2024 - link
In fact CAMM2 offers the best of both worlds.Silver5urfer - Tuesday, July 16, 2024 - link
Intel screwed up does not mean AMD's Arch also follow same. AMD's Zen 3 processors ran at 1.4V, Intel Sandy to SKL all ran at 1.4V. None of them exhibit the damn instability or the degradation to the point of no return.AMD needs to step up their TDP and wattage, the 9950X barely beats a 7950X. It's mediocre 41K CBR23, vs Zen 4's 38K. Compare to the Zen 3 vs Zen 4, the jump was over 14000points in CBR23.
AM5 socket is capped to 230W, no amount of PBO will exceed that. And PBO added Zen 4 goes to 40K and Zen 5 with Curve Shaper and PBO2 it will go upto 45K max.
AMD could have added more power, 50W for the R9, 25W for the R7 but they instead reduced it massively and also cut down the Base Clocks by a lot of MHz. Wonder if it's due to Intel's ARL dropping SMT / Hyperthreading.
evanh - Tuesday, July 16, 2024 - link
They are all unlocked. The power settings can be altered as you like.Terry_Craig - Tuesday, July 16, 2024 - link
Cinebench is irrelevant. In the real world Zen5 will beat it by larger margins.GeoffreyA - Monday, July 15, 2024 - link
Thanks for the article, Gavin. I am somewhat impressed by Zen 5, particularly what's going on in its floating-point domain and the hints of its encoding performance. Intel made an excellent move disabling AVX-512.Gavin Bonshor - Monday, July 15, 2024 - link
You're welcome. Zen 5 looks very promising, but we shall wait and see how it fares!GeoffreyA - Monday, July 15, 2024 - link
It could be slaughter!sharath.naik - Tuesday, July 16, 2024 - link
Intel will have the advantage starting with lunar lake in performance per watt. There is no getting around 3nm vs 4nm advantage. Also AMD famously does paper launches, they will launch it only to not find any products to buy. you still see there are a lot of laptops with AMD 7nm still being sold. It's their way of clearing the inventory.GeoffreyA - Tuesday, July 16, 2024 - link
Possible, but let's wait for the benchmarks and see.ballsystemlord - Monday, July 15, 2024 - link
AMD gives us AVX-512, while Intel keeps it disabled after pushing for it for years. Funny times.kpb321 - Tuesday, July 16, 2024 - link
Intel's problem is that they don't yet have AVX-512 on their E cores which makes things awkward. They theoretically could allow you to run it on the P cores but not the E cores but that has a lot of draw backs. You'd have to manage threads so they stayed on the P cores or they'd crash if they migrated to the E cores. You also might not get much benefit from it if you are limited to only P cores because you'd be loosing out on a lot of cores and the stuff that scales well on AVX-512 also tends to be the same type of things that scale well with more cores. Trying to run mixed code with AVX-512 on the P cores and run AVX-256 or something else on the E cores but that seems even more complicated. Intel does have a solution in the pipeline with AVX10 which will be available on the E cores eventually. The E cores will run it as combined 256 bit vectors so it will be slower but it will be able to run the same code.GeoffreyA - Tuesday, July 16, 2024 - link
The obvious solution was for them to implement AVX-512 on the E-cores, combining FPUs to do the job, and there wouldn't have been any problem. Apparently, AVX10 has issues of its own but I'm not too clued up on that at present.DanNeely - Wednesday, July 17, 2024 - link
That fixes the thread migration problem for single threaded workloads, but is only of limited help with multithreaded ones being a repeat of the old AMD shared FPU design.GeoffreyA - Thursday, July 18, 2024 - link
I hear what you're saying: run away from Bulldozer's approach. I meant it in the sense of combining the four 128-bit vector paths in a single Skymont core to make up 512 bits, similar to Zen 4's manner of joining two 256-bit paths. Granted, Crestmont had only three 128-bit paths, making it problematic.GeoffreyA - Tuesday, July 16, 2024 - link
It is funny, especially considering that fast AVX-512 will become one of the hallmarks of Zen 5. These instructions are being increasingly used in everyday applications. For example, the SVT-AV1 encoder and dav1d decoder. This perhaps explains why Handbrake's performance was so high.Slash3 - Monday, July 15, 2024 - link
Typo in headline, as an FYI.Slash3 - Monday, July 15, 2024 - link
(Also in the related page navigation drop-down)Gavin Bonshor - Monday, July 15, 2024 - link
Thank you!!! I haven't slept, so that one is on me. I appreciate itSlash3 - Monday, July 15, 2024 - link
Cheers, thanks! Appreciate all the hard work.kkilobyte - Monday, July 15, 2024 - link
Good and interesting, but what about the i9-14900KS test redo with Intel Default settings? You told us 2 months ago that you'd redo them, and I think the recent news about possible issues with this CPU make it even more relevant :"Gavin Bonshor - Friday, May 10, 2024 - link
Don't worry; I will be testing Intel Default settings, too. I'm testing over the weekend and adding them in."
So, will this promise be ever fullfilled?
Gavin Bonshor - Monday, July 15, 2024 - link
Email me please kkilobytekkilobyte - Monday, July 15, 2024 - link
Done :)kkilobyte - Friday, July 19, 2024 - link
Should I expect an answer from you someday? I didn't receive one so far...meacupla - Monday, July 15, 2024 - link
There's no point doing a retest of i9 13/14th gen right now. They're still failing after the patches have been applied.kkilobyte - Monday, July 15, 2024 - link
This is difficult to say, isn't it?I'm saying this because it seems to me that, for the most part, CPUs that got the patches applied received them rather late in their lifetime, having run under possibly "wrong" settigns for quite a while. So the mitigations in such a case may have come too late to prevent damage from happening.
I'm wondering if there are statistics about that so far. That would be interesting to see if applying such mitigations changed anything on the "lifetime without failure" of the affected CPU, when applied early on its lifetime.
Dante Verizon - Monday, July 15, 2024 - link
It seems that a 10% downclock solves the problem.Dante Verizon - Monday, July 15, 2024 - link
I wouldn't count on it. Look who owns this site.kkilobyte - Tuesday, July 16, 2024 - link
I still believe in the integrity of its journalists. Maybe I'm naive, but I think it is important, else what's the point of reading their articles?Dante Verizon - Thursday, July 18, 2024 - link
I don't doubt that the benchmark figures here are true, but I also believe that sites that need marketing funds can't be aggressive in pointing out manufacturers' problems.deil - Tuesday, July 16, 2024 - link
I feel like AI9HX chip will be massive win. I noticed asus put them in vivobooks, with decent otherwise specs, no extra trash gpu, 90W. It feels like the best thing you can get right now, especially thin and light, as "paired" with 4060 is a joke.integrated can run like lowest tdp 4060 mobile.
who in his right mind would want to double the power draw, for no other reasons than NVIDIA sticker and dedicated VRAM.
If I find any laptop with 2 or 3 slots for ssd, any format, I will take that, if not, I guess that VivoBook it is.
btw, It's really sad that small form laptops cannot find any space for a second m.2 slot.
just why. Especially with all those PCIE empty with no gpu....
nandnandnand - Tuesday, July 16, 2024 - link
Haven't checked the prices, but anything without a dGPU ought to be priced less. Unless we're talking Strix Halo next year.Terry_Craig - Tuesday, July 16, 2024 - link
And you get rid of bugs due to constant switching between dGPU and iGPU.HardwareDufus - Tuesday, July 16, 2024 - link
I would have liked to see an interim RDNA 3.5 for discrete cards too... Maybe a Navi 30 that would have powered the never released Radeon RX 7950 XTX. Might have been a nice stop gap. Maybe it was just too optimized for lower energy and mobile.RX 8000 built on RDNA 4 looks to target low/middle of the market around Q1/Q2 2025 with no high end part (assuming Navi 48 is low, Navi 44 is middle).
Zoolook13 - Wednesday, July 17, 2024 - link
What if it's just two sizes of chiplets, I haven't seen that AMD has said they are not targeting high end, we know they have been working on chiplet GPUs and now Nvidia is as well.StormyParis - Friday, July 19, 2024 - link
I'm on a i5-4570S for my main PC Nothing wrong with it really, except video deconding. Been hankering after an upgrade for a few years: Zen4, then Zen4-G... now it seems I've got to wait for Zen5 even probably Zen5-G but at some point I'll have to pull the trigger before MS takes Windows away from me ^^My one concern is USB reliability. I use USB a lot, and last time I dropped AMD because of USB issues. That was 11 yrs ago, but I've read in some places that USB still isn't rock-steady the way it is w/ Intel.