News AMD Ryzen AI 300 CPU beats Intel Core Ultra 200V CPU in Linux showdown — Strix Point was up to 1.6X faster than Lunar Lake

[Admin]

According to one review, Lunar Lake's performance in Linux disappoints, which found Ryzen AI 300 APUs to be much better.

AMD Ryzen AI 300 CPU beats Intel Core Ultra 200V CPU in Linux showdown — Strix Point was up to 1.6X faster than Lunar Lake : Read more

 

[TheSecondPower]

Phoronix's benchmark suite doesn't seem representative of ordinary use, especially for the lowest-power laptop market. In most of the web benchmarks—which do represent ordinary use—Phoronix's testing placed Lunar Lake as a stellar performer both in overall performance and power consumption. Much of the testing is multithreaded, which is certainly not where the 4/4+4/4 processor will win against the 4/8+8/16 processor. (big cores/big threads+little cores/little threads)

 

[redgarl]

Phoronix's benchmark suite doesn't seem representative of ordinary use, especially for the lowest-power laptop market. In most of the web benchmarks—which do represent ordinary use—Phoronix's testing placed Lunar Lake as a stellar performer both in overall performance and power consumption. Much of the testing is multithreaded, which is certainly not where the 4/4+4/4 processor will win against the 4/8+8/16 processor. (big cores/big threads+little cores/little threads)

Do I need to refer to this for you? REAL WORLD PERFORMANCE?

Stop Gaslighting...

View: https://youtu.be/Gj5_uc4ScNE?si=saqUN466-5_FwbQE

 

[blargh4]

What is the % of people buying thin+light laptops for heavy multithreaded workloads?

 

[Giroro]

Ryzen AI 9 HX 370 Is a 38 Watt (Up to 54W) chip designed for mid-to-large High-Performance / Desktop Replacement / Gaming laptops.
Core Ultra 7 256V is a 17W (which includes the RAM) high-efficiency chip designed for Ultra thin and Ultra portable laptops and meant to directly compete with Apple M3 and other ARM/RISC processors.
This is why I complain about how badly Intel the marketing of the Ultra X 2xxV series, or however we are supposed to genericize the absurd number of SKUs they made out of this single chip. Intel should have been selling this product in a way where people are impressed that they were able to achieve reasonably close performance at half the power. But instead we are still talking about how it loses in absolute performance against a totally different market segment where they should not be trying to compete.

Intel is still using the marketing tactics of a stable, growing monopoly in an era where they are bleeding market share and just had to lay off 15% of the company. This is what happens when you fire low level staff instead of the overpriced C-suite execs who got your company into this mess. The results continue to speak for themselves.

 

[rtoaht]

Do I need to refer to this for you? REAL WORLD PERFORMANCE?

Stop Gaslighting...

View: https://youtu.be/Gj5_uc4ScNE?si=saqUN466-5_FwbQE

So, if my AI 300 laptop is slower than my friend's Core Ultra 200V laptops and doesn't last as long on battery should I just watch that video and be happy?

 

[bit_user]

Phoronix did note that the 265V was at least power efficient and did well in single-threaded benchmarks, but it lost primarily when workloads could utilize more threads.

Obviously, the multithreaded performance was going to suck. And, about that single-threaded performance...

In most of the web benchmarks—which do represent ordinary use—Phoronix's testing placed Lunar Lake as a stellar performer both in overall performance and power consumption.

You're not looking at them critically enough. If you pay attention, those benchmarks not only had the Ryzen AI machines performing abysmally, they also had the Ryzen AI using significantly less power than the Lunar Lake. So, something was obviously wrong.

It turns out that the Ryzen AI CPUs were running the lightly-threaded jobs on the Zen 5C cores, due to a known Linux scheduling bug affecting them. So, they were at a substantial disadvantage and much worse than what we've seen from comparable benchmarks on Windows.

In fact Notebook Check found Ryzen AI beat the Ultra 7 258V on all of the the web benchmarks they ran:

https://www.notebookcheck.net/Intel...its-everyday-efficiency-is-good.893405.0.html
​

The 288V managed to take some wins on the WebXPRT benchmarks, but that's not the model Phoronix tested (and the 288V has worse power efficiency, not surprisingly).

 

[bit_user]

What is the % of people buying thin+light laptops for heavy multithreaded workloads?

Developers, developers, developers, developers. Developers! We want a thin & light laptop, but we want fast compiles when we do them.

Everyone seems to think you're either doing long-running renders (in which case something like a workstation definitely makes sense), or else you don't care about MT performance. This is a false dichotomy.

Apple seems to think there's a market for this.

 

[rluker5]

Developers, developers, developers, developers. Developers! We want a thin & light laptop, but we want fast compiles when we do them.

So 1%?

 

[bit_user]

So 1%?

LOL. I have no idea, to be honest.

 

[rluker5]

That makes 2 of us then.

 

[TheHerald]

What is the % of people buying thin+light laptops for heavy multithreaded workloads?

All of them apparently. They don't care about battery life, just running cinebench on a loop. Till of course Intel wins in MT performance, then thin and light laptops will be about battery life again.

So, if my AI 300 laptop is slower than my friend's Core Ultra 200V laptops and doesn't last as long on battery should I just watch that video and be happy?

Yes, but make sure your AI 300 laptop is plugged in cause you won't make it through the video on battery.

 

[TheSecondPower]

You're not looking at them critically enough. If you pay attention, those benchmarks not only had the Ryzen AI machines performing abysmally, they also had the Ryzen AI using significantly less power than the Lunar Lake. So, something was obviously wrong.

I didn't say that the Core 200V chip beat the Ryzen 300 chips, I said it was a stellar performer at web tasks. It's a way ahead of previous Intel generations and should be in the ballpark of the Ryzen 300 chips in most web tasks. JavaScript apps are usually a single thread, even when doing asynchronous work.

 

[bit_user]

... in most web tasks. JavaScript apps are usually a single thread, even when doing asynchronous work.

I'm probably not telling you anything you don't know, but some of the web benchmarks are definitely multi-threaded. I'd imagine that's the main reason the Ryzen AI processors did so well in Notebook Check's web benchmarks. In terms of real-world user experience, I'd agree that lightly-threaded performance is what's most relevant to web usage.

Somewhere around 7+ years ago, I saw an article on Anandtech that did some analysis of web performance, analyzing multi-core scaling of different web browsers. That wasn't the main point of the piece, but they devoted like a page to it. It does seem web browsers had some internal multithreading to speed up things like layout, image decoding, rendering, fetching, JIT, etc. In practical usage, I expect time is still probably dominated by a minority of threads, but it was interesting to see how browsers had evolved to take advantage of more cores, as 8+ threads was quickly becoming the norm in both desktop and phone platforms.

 

[bit_user]

All of them apparently. They don't care about battery life, just running cinebench on a loop. Till of course Intel wins in MT performance, then thin and light laptops will be about battery life again.

I think you're confused, here. Just because an article tests cinebench or someone mentions MT performance & efficiency doesn't mean anyone thinks it's a top priority for these products. However, you also can't dismiss them as irrelevant, because there are clearly some users out there who want/need it.

I'll give you another use case, which is important for a lot of corporate users. At work, I mostly use an Intel i5-1250P, which has 4P + 8E cores (16 threads). When I login and fire up task manager, nearly all cores/threads are saturated for about the first minute, before it starts to settledown. That's because there's a ton of stuff that, whether it makes sense or not, wants to run immediately upon login. Performance of everything is further bogged down by not only anti-virus, but other sorts of security software.

Where it becomes really painful is when I need to join a meeting on MS Teams. I wakeup my laptop to find it had rebooted to install some updates, so I have to login instead of just unlocking it. I can do nothing but sit there and wait, for at least 30 seconds (which feels like an eternity, if I'm already late to a meeting) before I can finally join the call. And I'm quoting you performance numbers when it's actually on A/C power and plenty of clearance around the cooling vents + ducts (plus no dust or lint clogging it).

Now, if you give someone a new notebook and it doesn't improve on this important metric, I'm sure most of us would be disappointed. What makes it worse is that software tends to get slower with time, not faster. So, you could imagine that what might be 30-40 seconds today could eventually stretch beyond a minute, before they eventually get another upgrade.

In designing a 4+4 processor, Intel seems to have focused on the Apple M3. However, I'll bet most MacOS installs don't have so much bloatware. Plus, Apple lets you go from a 2.7 pound Macbook Air to a 3.5 pound Macbook Pro, if you need more performance. That still classifies as "thin & light" in my book. It's certainly thinner & lighter than my i5-1250P laptop.

 

[TheHerald]

I think you're confused, here. Just because an article tests cinebench or someone mentions MT performance & efficiency doesn't mean anyone thinks it's a top priority for these products. However, you also can't dismiss them as irrelevant, because there are clearly some users out there who want/need it.

I'll give you another use case, which is important for a lot of corporate users. At work, I mostly use an Intel i5-1250P, which has 4P + 8E cores (16 threads). When I login and fire up task manager, nearly all cores/threads are saturated for about the first minute, before it starts to settledown. That's because there's a ton of stuff that, whether it makes sense or not, wants to run immediately upon login. Performance of everything is further bogged down by not only anti-virus, but other sorts of security software.

Where it becomes really painful is when I need to join a meeting on MS Teams. I wakeup my laptop to find it had rebooted to install some updates, so I have to login instead of just unlocking it. I can do nothing but sit there and wait, for at least 30 seconds (which feels like an eternity, if I'm already late to a meeting) before I can finally join the call. And I'm quoting you performance numbers when it's actually on A/C power and plenty of clearance around the cooling vents + ducts (plus no dust or lint clogging it).

Now, if you give someone a new notebook and it doesn't improve on this important metric, I'm sure most of us would be disappointed. What makes it worse is that software tends to get slower with time, not faster. So, you could imagine that what might be 30-40 seconds today could eventually stretch beyond a minute, before they eventually get another upgrade.

In designing a 4+4 processor, Intel seems to have focused on the Apple M3. However, I'll bet most MacOS installs don't have so much bloatware. Plus, Apple lets you go from a 2.7 pound Macbook Air to a 3.5 pound Macbook Pro, if you need more performance. That still classifies as "thin & light" in my book. It's certainly thinner & lighter than my i5-1250P laptop.

1) If you give someone a new laptop, it will improve on those "important" metrics, because companies usually don't renew their fleet in 2 years.

2) Your usecase is basically like a desktop. Which is fine but again you are missing the point of a thin and light laptop

3) I can come up with some scenarios as well that ends up at having the battery life to join that teams call is better than installing the updates fast but you end up without battery.

In fact, i will do just that. Every couple of months me and my missus grab our camper and go for vacation. Usually we go to the beach with our laptops and work there. Sadly if we run out of battery (which we always do cause we cant get 8 hours of work time) we have to go back to the camper to charge. Giving me a laptop that improves on this important metric, oh boy.

EG. Macs are a different OS. Most clients I have that changed to macbooks threw them away and went back to windows within months. Obviously it's a PEBKAC issue,they didn't know how to use it I guess, but the point stands, I don't see them as windows competitors.

 

[nogaard777]

Chip designed for ultra low power applications doesn't perform as well as chip with more cores designed for higher power uses and thicker chassis? Shocker!

 

[blargh4]

It's a world of tradeoffs out here. If I was getting a laptop for work, I definitely wouldn't want a macbook air-class product. But if I want a Macbook Air-like PC, heaps of multithreaded perf wouldn't be a plus for what I value about that class of machine - you'd need beefier cooling to sustain it and probably a bigger battery. Battery life for low CPU duty cycle activities like web browsing, on the other hand, is a big plus, and Lunar Lake seems to be a major step forward for x86 designs.

 

[watzupken]

I think the truth is everyone wants a very performant laptop that is light and comes with great battery life. Which is why when Apple pulled it off with the initial M1 SOC, these ARM based MacBook were selling really well. Then we have the recent Qualcomm Elite SOCs that also generated a lot of interest. Intel's Lunar Lake clearly traded all round performance for efficiency, which is why it may appeal to most, but not to some, depending on your workflow. Moreover, you also have the limitation of 32GB RAM that is on die, and you can't add more. Again, removing a certain market from being interested in it. I guess Intel figured that as long as overall Lunar Lake sales is good, they are willing to give up certain market segments.

What I found interesting with Lunar Lake is the Battlemage iGPU. It just shows how little attention AMD gave to their GPU segment, that they basically ceded their iGPU dominance to Intel in just 1 generation. The TSMC 3nm certainly gave Intel and edge here, but I am not sure by how much. But I do observed that AMD hardly made any progress with between RDNA 2, 3 and 3.5.

 

[bit_user]

1) If you give someone a new laptop, it will improve on those "important" metrics, because companies usually don't renew their fleet in 2 years.

My company leases these machines for 3 years. I got this one in Summer of 2022. That means it's likely to be replaced with a Lunar Lake, unless I can convince them to give me an Arrow Lake, instead.

2) Your usecase is basically like a desktop. Which is fine but again you are missing the point of a thin and light laptop

Huh? Literally every single employee in my company has this same security software installed on their Windows machines, whether it's a desktop, laptop, etc. I think you can get a Mac, but you need to have a business reason to use it instead of Windows.

In fact, i will do just that. Every couple of months me and my missus grab our camper and go for vacation. Usually we go to the beach with our laptops and work there. Sadly if we run out of battery (which we always do cause we cant get 8 hours of work time) we have to go back to the camper to charge. Giving me a laptop that improves on this important metric, oh boy.

That's nice. I usually find the idea of working outside is nicer than the reality, but I don't dismiss those use cases people have where you actually need decent battery life. The few occasions I've used my laptop in places where I couldn't plug into A/C power, I've found the battery life to be atrocious, even for fairly light-duty tasks, but I should note that the machine is rarely actually idle, due to all the various security and spyware loaded on this thing. So, if I do get a Lunar Lake as a replacement, at least that bit will be a nice change.

 

[bit_user]

Chip designed for ultra low power applications doesn't perform as well as chip with more cores designed for higher power uses and thicker chassis? Shocker!

Notebook Check found the following Cinebench 24 performance, when several different laptop CPUs were each configured at the same power level:

TDP	Intel Core Ultra 7 258V	Ryzen AI 9 HX 370	Ryzen AI 9 365	Intel Core Ultra 7 155H
15 watts	445 points	672 points	590 points	323 points
20 watts	512 points	767 points	683 points	433 points
28 watts	587 points	876 points	787 points	573 points

Source: https://www.notebookcheck.net/Intel...its-everyday-efficiency-is-good.893405.0.html
​

So, both Ryzen AI CPUs were faster at 15 W than either Intel CPUs at 28 W! The victory over Lunar lake is obviously due to more cores/threads, but it still shows that TDP isn't everything, nor is Lunar Lake's process node advantage.

Perhaps more interesting is Meteor Lake's performance, since it has 16 cores & 22 threads, whereas the Ryzen AI 9 365 has only 10 cores and 20 threads. Both are on allegedly the same tier of node, with Meteor Lake using Intel 4 and Ryzen AI using TSMC N4P.

 

[bit_user]

What I found interesting with Lunar Lake is the Battlemage iGPU. It just shows how little attention AMD gave to their GPU segment, that they basically ceded their iGPU dominance to Intel in just 1 generation. The TSMC 3nm certainly gave Intel and edge here, but I am not sure by how much. But I do observed that AMD hardly made any progress with between RDNA 2, 3 and 3.5.

Wait until AMD's Strix Halo, launching early next year. That will feature a substantially larger iGPU and 256-bit on-package memory. Early leaks suggests GPU performance better than a PS5. Still not on par with the fastest Apple laptops, which feature a 512-bit memory interface, but definitely progress.

 

[DS426]

Ryzen AI 9 HX 370 Is a 38 Watt (Up to 54W) chip designed for mid-to-large High-Performance / Desktop Replacement / Gaming laptops.
Core Ultra 7 256V is a 17W (which includes the RAM) high-efficiency chip designed for Ultra thin and Ultra portable laptops and meant to directly compete with Apple M3 and other ARM/RISC processors.
This is why I complain about how badly Intel the marketing of the Ultra X 2xxV series, or however we are supposed to genericize the absurd number of SKUs they made out of this single chip. Intel should have been selling this product in a way where people are impressed that they were able to achieve reasonably close performance at half the power. But instead we are still talking about how it loses in absolute performance against a totally different market segment where they should not be trying to compete.

Intel is still using the marketing tactics of a stable, growing monopoly in an era where they are bleeding market share and just had to lay off 15% of the company. This is what happens when you fire low level staff instead of the overpriced C-suite execs who got your company into this mess. The results continue to speak for themselves.

Right, though being fair to Intel, it was always going to be an uphill marketing battle for Intel as absolute performance is the sexiest metric for most observers in the PC world, even if the product is supposed to be clearly targeted for ultrabooks meant to compete with Apple Silicon and ARM ultrabooks. Then again, battery life is a sexy spec for most notebook buyers, so they do regain some traction.

I agree though that they need to get their s**t sorted out by the time Arrow Lake launches.