So what are the power numbers for this chip? Sure anything can beat M series chips if it uses more power.
As mentioned in the article, the 7840u was designed for “thin and light” notebooks, ergo machines comparable to the MacBook Air (only with a fan or two). We won’t know stuff like power consumption and fan noise until this chip goes into an actual production model, but AMD is already writing cheques it may not be able to cash in terms of claims about potential performance, because some factors in any real-world comparison are going to rely on machines built by other companies.
In that case, a MacBook Pro with m2 max is more comparable, since it’s a thin and light like a MacBook Air “only with a fan or two.”
AMD wants to compare down, combining their CPUs with higher end GPUs, but really its competitor is the m2 max, which uses even less power.
Thanks for this, very interesting. Assuming there is no other graphics assistance in the ROG Ally than just what's on the Z1 Extreme, quiet mode is very comparable to the (regular) M2, except that instead of being slower, the M2 is faster ... AND still uses a bit less power.
I really wish people would stop using Cinebench for CPU comparisons.
Firstly, it leans heavily on the SIMD instruction set which very few other applications do.
Secondly, it’s RT library is Embree had Apple 2nd Neon SIMD disabled by its GitHub custodian - Intel (see GitHub PR330). Go figure why it doesn’t paint M-series in a good light.
Clearly it’s time for the Wintel platform to be re-architected. The X86 architecture has been around for 40 plus years and despite dramatic advances in chip technology, performance of Windows-based machines will be forever hamstrung until a fresh architecture is implemented. If nothing else, Apple has shown there are ways (such as using Rosetta) to migrate to new chip architectures. Though in the Apple ecosystem it is a much simpler challenge than in the Wintel world. Apple has also shown just how poorly managed Intel has been over the past ten years. Fat, dumb and happy with their market dominance, they just let their marketing department drive the pace of incremental performance changes over time. Not a lot of significant design changes. Just build systems with bigger power supplies and fans and you could support performance improvements through the use of higher clock rates and more transistors. Just look at how long it took Intel to transition off 14 nm process. Intel’s stock price reflects the disappointing performance of the executives running the company. Apple’s future is highly dependent on TSMC. TSMC is at risk of political upheaval in Taiwan. The sooner TSMC gets facilities built outside Taiwan, the better for Apple. The same goes for Foxconn and China.
Thanks for this, very interesting. Assuming there is no other graphics assistance in the ROG Ally than just what's on the Z1 Extreme, quiet mode is very comparable to the (regular) M2, except that instead of being slower, the M2 is faster ... AND still uses a bit less power.
I have no idea what you mean by "no other graphics assistance in the ROG Ally", the only chip in there is the Z1 Extreme and this is a CPU test.
I also don't know how you came to the conclusion that the M2 is faster "AND" uses less power. The Points / Watt is higher for the Z1 Extreme in that test (69 > 64).
Please go back and read my original post (number 40) again.
So what are the power numbers for this chip? Sure anything can beat M series chips if it uses more power.
As mentioned in the article, the 7840u was designed for “thin and light” notebooks, ergo machines comparable to the MacBook Air (only with a fan or two). We won’t know stuff like power consumption and fan noise until this chip goes into an actual production model, but AMD is already writing cheques it may not be able to cash in terms of claims about potential performance, because some factors in any real-world comparison are going to rely on machines built by other companies.
Well Apple had a thin and light MacBook line up on x86 CPU's so that doesn't meant shit. They still spun the fans up on max all the time and only had decent battery life. My point was, what they said doesn't mean shit. It's all talk with nothing to back it up.
I also don't know how you came to the conclusion that the M2 is faster "AND" uses less power. The Points / Watt is higher for the Z1 Extreme in that test (69 > 64).
Please go back and read my original post (number 40) again.
Sorry, I did misread it. Points/Watt higher = better. My bad.
So what are the power numbers for this chip? Sure anything can beat M series chips if it uses more power.
As mentioned in the article, the 7840u was designed for “thin and light” notebooks, ergo machines comparable to the MacBook Air (only with a fan or two). We won’t know stuff like power consumption and fan noise until this chip goes into an actual production model, but AMD is already writing cheques it may not be able to cash in terms of claims about potential performance, because some factors in any real-world comparison are going to rely on machines built by other companies.
And yet that still doesn’t answer the question about power numbers.
We’ve got Intel chips meant for thin and light notebooks that melt them down.
Actual, factual hard numbers is what we need to see.
So what are the power numbers for this chip? Sure anything can beat M series chips if it uses more power.
As mentioned in the article, the 7840u was designed for “thin and light” notebooks, ergo machines comparable to the MacBook Air (only with a fan or two). We won’t know stuff like power consumption and fan noise until this chip goes into an actual production model, but AMD is already writing cheques it may not be able to cash in terms of claims about potential performance, because some factors in any real-world comparison are going to rely on machines built by other companies.
And yet that still doesn’t answer the question about power numbers.
We’ve got Intel chips meant for thin and light notebooks that melt them down.
Actual, factual hard numbers is what we need to see.
Whether this chip is in a handheld, laptop or mini desktop, it has a lot of potential.
I'd also note that the ASUS ROG Ally Z1 Extreme is only $699 USD with 16 GB LPDDR5-6400 RAM and 512 GB NVMe PCIe 4.0 SSD. So prices for devices with these chips appear to be quite reasonable.
Clearly it’s time for the Wintel platform to be re-architected. The X86 architecture has been around for 40 plus years and despite dramatic advances in chip technology, performance of Windows-based machines will be forever hamstrung until a fresh architecture is implemented. If nothing else, Apple has shown there are ways (such as using Rosetta) to migrate to new chip architectures. Though in the Apple ecosystem it is a much simpler challenge than in the Wintel world. Apple has also shown just how poorly managed Intel has been over the past ten years. Fat, dumb and happy with their market dominance, they just let their marketing department drive the pace of incremental performance changes over time. Not a lot of significant design changes. Just build systems with bigger power supplies and fans and you could support performance improvements through the use of higher clock rates and more transistors. Just look at how long it took Intel to transition off 14 nm process. Intel’s stock price reflects the disappointing performance of the executives running the company. Apple’s future is highly dependent on TSMC. TSMC is at risk of political upheaval in Taiwan. The sooner TSMC gets facilities built outside Taiwan, the better for Apple. The same goes for Foxconn and China.
Wintel x86 clearly still has some life left in it. But apple has signaled the beginning of the end. AMD and INTEL are wringing the last bit of juice out of it. But soon, they’ll have to go RISC to avoid put air conditioners inside computer cases.
The GPU battle is actually a similar story. NVIDIA and AMD can get some impressive performance out of their architectures - but they also need a ton of juice and cooling to get the really impressive numbers. That can hold them for while though simply because apple hasn’t matched nor exceeded the high end GPUs yet. But they will before too long. And that’s going to get very interesting. It’s not out of the question for apple to partner with Nvidia to integrate their high end GPUs into desktop apple silicon. But Apple is a proud company. They’d rather beat someone than join them. And they have learned better than to pump up a competitor with too much business (see Samsung).
So what are the power numbers for this chip? Sure anything can beat M series chips if it uses more power.
As mentioned in the article, the 7840u was designed for “thin and light” notebooks, ergo machines comparable to the MacBook Air (only with a fan or two). We won’t know stuff like power consumption and fan noise until this chip goes into an actual production model, but AMD is already writing cheques it may not be able to cash in terms of claims about potential performance, because some factors in any real-world comparison are going to rely on machines built by other companies.
And yet that still doesn’t answer the question about power numbers.
We’ve got Intel chips meant for thin and light notebooks that melt them down.
Actual, factual hard numbers is what we need to see.
Whether this chip is in a handheld, laptop or mini desktop, it has a lot of potential.
I'd also note that the ASUS ROG Ally Z1 Extreme is only $699 USD with 16 GB LPDDR5-6400 RAM and 512 GB NVMe PCIe 4.0 SSD. So prices for devices with these chips appear to be quite reasonable.
Not doing extrapolations dude. I have no doubt it’s impressive. So are Intels best chips.
I meant the actual numbers of the SOCs in question as they ran throughout the “benchmarks.” Actual power and thermals - while generating publicized performance.
actual numbers for the actual part. Not you math project.
and the world knows benchmarks are only a small part of the story and can be manipulated by less honest companies. Lots of testing has been done with apple silicon compared to processors that beat it in benchmarks only to see Apple Solicon win out in real world usage.
The Apple M-series have shown better against other processor than they should on multithrreded benchmarks, and this article reveals just that. The difference is mutithreaded cores, what Intel calls hyperthreading, as stated in the article. What you don't point out is why Cinebench shows so much better than artificial benchmarks like Geekbench. The difference is that Cinebench will schedule more threads than cores in order to exploit multithreading. The artificial benchmarks allocate one thread per core.
There is no guarantee that exploiting on-chip multithreading is a performance win, it depends on the task. But for many multithreaded applications, it's a big win. The artificial benchmarks really should benchmarks separately with N and 2N threads, or something similar.
I'm confused. This is comparing against the 10-core M2, not the Pro or the Max variants?
Can this new AMD processor be seamlessly united with a second processor to increase the performance like Apple has for the M-series?
All AMD processors are based on chiplets. So the answer is "yes", though in a far more elegant way than Apple. So far, Apple's able to unite two processors. The Pro is a different slice of silicon than the plain M1/M2, albeit with the same kind of cores. The Max is two Pros stuck together. The rumored Ultra does exist, but would have four Pro chips.
AMD builds CPU chiplets with 8 cores. So their 64 and 128 core chips -- obviously not a laptop thing -- include 8 and 16 chiplets, respectively. Way more scaleable. However, they are only recently getting into GPU chiplets. So matching Apple's doubled up GPU/NPU would require a different chip today, but probably not for long. AMD is on their third generation of interconnect architecture.
So what are the power numbers for this chip? Sure anything can beat M series chips if it uses more power.
As mentioned in the article, the 7840u was designed for “thin and light” notebooks, ergo machines comparable to the MacBook Air (only with a fan or two). We won’t know stuff like power consumption and fan noise until this chip goes into an actual production model, but AMD is already writing cheques it may not be able to cash in terms of claims about potential performance, because some factors in any real-world comparison are going to rely on machines built by other companies.
And yet that still doesn’t answer the question about power numbers.
We’ve got Intel chips meant for thin and light notebooks that melt them down.
Actual, factual hard numbers is what we need to see.
Whether this chip is in a handheld, laptop or mini desktop, it has a lot of potential.
I'd also note that the ASUS ROG Ally Z1 Extreme is only $699 USD with 16 GB LPDDR5-6400 RAM and 512 GB NVMe PCIe 4.0 SSD. So prices for devices with these chips appear to be quite reasonable.
Not doing extrapolations dude. I have no doubt it’s impressive. So are Intels best chips.
I meant the actual numbers of the SOCs in question as they ran throughout the “benchmarks.” Actual power and thermals - while generating publicized performance.
actual numbers for the actual part. Not you math project.
and the world knows benchmarks are only a small part of the story and can be manipulated by less honest companies. Lots of testing has been done with apple silicon compared to processors that beat it in benchmarks only to see Apple Solicon win out in real world usage.
If you would take a look, this is a pretty extensive review with quite a few details on the actual performance of the Z1 Extreme (essentially the 7840u). That being said, if you have difficulty understanding what I highlighted in my original post, you're going to get lost pretty quickly. At least the information is there for other users to see.
So what are the power numbers for this chip? Sure anything can beat M series chips if it uses more power.
As mentioned in the article, the 7840u was designed for “thin and light” notebooks, ergo machines comparable to the MacBook Air (only with a fan or two). We won’t know stuff like power consumption and fan noise until this chip goes into an actual production model, but AMD is already writing cheques it may not be able to cash in terms of claims about potential performance, because some factors in any real-world comparison are going to rely on machines built by other companies.
And yet that still doesn’t answer the question about power numbers.
We’ve got Intel chips meant for thin and light notebooks that melt them down.
Actual, factual hard numbers is what we need to see.
Whether this chip is in a handheld, laptop or mini desktop, it has a lot of potential.
I'd also note that the ASUS ROG Ally Z1 Extreme is only $699 USD with 16 GB LPDDR5-6400 RAM and 512 GB NVMe PCIe 4.0 SSD. So prices for devices with these chips appear to be quite reasonable.
Not doing extrapolations dude. I have no doubt it’s impressive. So are Intels best chips.
I meant the actual numbers of the SOCs in question as they ran throughout the “benchmarks.” Actual power and thermals - while generating publicized performance.
actual numbers for the actual part. Not you math project.
and the world knows benchmarks are only a small part of the story and can be manipulated by less honest companies. Lots of testing has been done with apple silicon compared to processors that beat it in benchmarks only to see Apple Solicon win out in real world usage.
If you would take a look, this is a pretty extensive review with quite a few details on the actual performance of the Z1 Extreme (essentially the 7840u). That being said, if you have difficulty understanding what I highlighted in my original post, you're going to get lost pretty quickly. At least the information is there for other users to see.
I looked. You didn’t read. Your math project doesn’t tell the story. It’s an extrapolation. Let’s see the actual numbers that were required to generate performance throughout the run. And then let’s see it done in the real world.
I expect the AMD unit will wining key area. But not more performance with more efficiency - outside of any specific hardware addition to more easily pass a desired benchmark.
Comments
Firstly, it leans heavily on the SIMD instruction set which very few other applications do.
So please stop using Cinebench.
I also don't know how you came to the conclusion that the M2 is faster "AND" uses less power. The Points / Watt is higher for the Z1 Extreme in that test (69 > 64).
Please go back and read my original post (number 40) again.
This is an impressive chip.
Red = Turbo
Green = Performance
Whether this chip is in a handheld, laptop or mini desktop, it has a lot of potential.
I'd also note that the ASUS ROG Ally Z1 Extreme is only $699 USD with 16 GB LPDDR5-6400 RAM and 512 GB NVMe PCIe 4.0 SSD. So prices for devices with these chips appear to be quite reasonable.
actual numbers for the actual part. Not you math project.
and the world knows benchmarks are only a small part of the story and can be manipulated by less honest companies. Lots of testing has been done with apple silicon compared to processors that beat it in benchmarks only to see Apple Solicon win out in real world usage.
There is no guarantee that exploiting on-chip multithreading is a performance win, it depends on the task. But for many multithreaded applications, it's a big win. The artificial benchmarks really should benchmarks separately with N and 2N threads, or something similar.
AMD builds CPU chiplets with 8 cores. So their 64 and 128 core chips -- obviously not a laptop thing -- include 8 and 16 chiplets, respectively. Way more scaleable. However, they are only recently getting into GPU chiplets. So matching Apple's doubled up GPU/NPU would require a different chip today, but probably not for long. AMD is on their third generation of interconnect architecture.