No idea. But I have a strong suspicion that at least Radeon results are max fps
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Intel Alder Lake vs. Apple M1
- Thread starter leman
- Start date
- Sort by reaction score
No idea. But I have a strong suspicion that at least Radeon results are max fps
I'm talking about computing devices excluding phones, not just laptops. I think that the vast majority are Chromebooks these days.
Yeah the numbers will be checked by reviewers (like Gamers Nexus). Usually AMD isn’t bold enough to include numbers (that can be checked) if they are going to lie.
Chromebooks are not that common. They sold 37 million units in 2021, out of 250-300 million laptops sold.
More Chromebooks than Macs were sold in 2021 then
2021 Mac shipments grew twice as fast as overall PC shipments - 9to5Mac
A new report shows that 2021 Mac shipments grew almost twice as fast as the PC market as a whole. Annual growth for Mac was 28.3% ...
9to5mac.com
The hard core gamers will get a 12900KS or 12900K, not the 12900. People like me that need "good" performance will get the 12900. (of course, the hardcore laptop gamers will get the 12900HS, but I would worry about cooling if I were them!)
What's surprising to me is the performance they're getting about the real consumer chips, the i7 and i5... (mobile version or not) They really bumped up the performance in the last couple of gens.
It's too hard to get threadripper motherboards oddly -- I know a guy at work that has been wanting one but can't get it. He's hard core AMD and doesn't like intel particularly. He has one of my old i5 Mac Mini's controlling his 3d printer.
More than 10% is actually impressive to me. I doubt if Apple sells that many.
What matters to many Mac users are either artists, video makers and picture takers programs too! They want HDR tinkering too! So talk about frames unless it's about video frames in HDR shot! Most picture takes like raw pictures too!
2021 Mac shipments grew twice as fast as overall PC shipments - 9to5Mac
A new report shows that 2021 Mac shipments grew almost twice as fast as the PC market as a whole. Annual growth for Mac was 28.3% ...
9to5mac.com
Apple sold 28,958,000 Macs in 2021. The cheapest Mac is $699 and cheapest Macbook is $999-$899.
That's really good sales considering Chromebooks are MUCH cheaper. Apple will make more profit.
I'm actually not a fan of chromebooks, nor do I really care about hardware profits of anyone -- but I do own Windows and Mac computers.
The thing about this **** is that while Rembrandt/Zen 3+/Ryzen 6000 would probably be my first choice in a laptop (barely over ADL-P 2+8, since Xe sucks, though close battle there IMO depending on binning) on the boomer ISA - it's absolutely pathetic how high either of them have to clock and how much cache has been devoted to..
A Geekbench 5 ST index of about 1571 for the 6900HS. That's not bad, and fantastic in isolation. Of course, here comes the lede (and really it's expected)
It does so @ 4.86GHz, and of course has ~ 20MB of L3 cache dedicated to the CPU unlike an SLC cache. (Also you can evaluate the actual values during the benchmark instead of the listed maximum, provided they didn't screw with reporting, by appending .gb5 to a web GB result btw).
Probably consumes 14-18W for a core at those frequencies, and I may have slightly lowballed really. The X86 volunteer defense caucus (PC gamers, people with conservative (lowercase c) impulses that they cannot rid themself of may wishcast about how "it's designed to clock higher" but that's not necessarily an instrinic part of the microarchitecture. I mean, it is in that you'd change your design depending on the goal — but if Apple were to implement and fabricate the M1 on predominantly higher performance libraries and reap higher clock frequencies, I imagine the main reason wide/large cores are "said to struggle" or "be difficult" with higher clocks is simply that you increase the odds of having some especially slower critical path in the core limiitng you by virtue of straightforwardly increasing the core's logical area (irrespective of library), or just the issue with e.g. ring buses and cache speeds.
Still, this is all exaggerated in that in principle Apple could presumably build an M1 designed to clock higher with minimal architectural differences, even at 3.8-4.4GHz, I think the cope would be priceless. Really, a part of me wishes they would do it in demonstration lol (they never would, but still)
A Geekbench 5 ST index of about 1571 for the 6900HS. That's not bad, and fantastic in isolation. Of course, here comes the lede (and really it's expected)
It does so @ 4.86GHz, and of course has ~ 20MB of L3 cache dedicated to the CPU unlike an SLC cache. (Also you can evaluate the actual values during the benchmark instead of the listed maximum, provided they didn't screw with reporting, by appending .gb5 to a web GB result btw).
Probably consumes 14-18W for a core at those frequencies, and I may have slightly lowballed really. The X86 volunteer defense caucus (PC gamers, people with conservative (lowercase c) impulses that they cannot rid themself of may wishcast about how "it's designed to clock higher" but that's not necessarily an instrinic part of the microarchitecture. I mean, it is in that you'd change your design depending on the goal — but if Apple were to implement and fabricate the M1 on predominantly higher performance libraries and reap higher clock frequencies, I imagine the main reason wide/large cores are "said to struggle" or "be difficult" with higher clocks is simply that you increase the odds of having some especially slower critical path in the core limiitng you by virtue of straightforwardly increasing the core's logical area (irrespective of library), or just the issue with e.g. ring buses and cache speeds.
Still, this is all exaggerated in that in principle Apple could presumably build an M1 designed to clock higher with minimal architectural differences, even at 3.8-4.4GHz, I think the cope would be priceless. Really, a part of me wishes they would do it in demonstration lol (they never would, but still)
Slowly seeing some ARM laptops getting released
Lenovo's Arm-Powered ThinkPad Touts 28-Hour Battery Life
https://www.tomshardware.com/news/lenovo-thinkpad-x13s-arm-laptopI’m actually quite pleased if Apple simply doesn’t go there. I agree that this has nothing to do with “can’t”, it’s just a decision they have (so far) made. But buying ever smaller increases in performace by ever increasing power draw seems like a dead end for the industry. It’s safe, in a sense, but it creates problems that people simply shouldn’t have to deal with. Increasing cost/power draw/noise for a modest increase in performance is a bad deal for most consumers and can only hurt uptake in the long run.
We’ll see what the personal computer market looks like post Covid.
Currently, AMD for instance simply doesn’t want to sell their APUs to selfbuilders, nor does anyone offer something like an XBox sx to PC gamers. I wonder if this might change if the market contracts, or if intel/AMD will double down on segmenting the market artificially in an attempt to keep margins up (at the cost of volume).
Isn't that what Microsoft is doing? Or are you saying no one is offering a APU with the power of a XSX for the price of the XSX with Windows included? (Which I think are two different things, though not everyone may agree)
They kind of do sell it .
AMD 4700S Review: Defective PlayStation 5 Chips Resurrected
Break out the broken chips
AMD 4700S 8-Core Processor Desktop Kit
AMD 4700S 8-Core Processor Desktop Kit powered by AMD “Zen 2” architecture delivers incredible performance for multitasking and heavy workloads. Learn more!
the iGPU doesn’t work . If there wasn’t an Xbox/PS4 shortage that would be an indicator AMD was main source of blocking. If the GPU works at this point it should go toward reducing shortfalls.[ Microsoft and Sony have some ecific elements so not entirely AMD to sell. AMD is probably buying these defects off owners to lower their overhead costs ]
The consoles and most laptops are going to have BGA packages. It makes little sense to sell those to random people. For BGA context, selling boards with the CPU and likely memory attached . That runs counter to “self builders” who are more focused on control than “build” .
For the sockets they do also . For example AM4 socket:
AMD Ryzen 7 5700G - Ryzen 7 5000 G-Series Cezanne (Zen 3) 8-Core 3.8 GHz Socket AM4 65W AMD Radeon Graphics Desktop Processor - 100-100000263BOX - Newegg.com
Buy AMD Ryzen 7 5700G - Ryzen 7 5000 G-Series Cezanne (Zen 3) 8-Core 3.8 GHz Socket AM4 65W AMD Radeon Graphics Desktop Processor - 100-100000263BOX with fast shipping and top-rated customer service. Once you know, you Newegg!
pragmatically for a ‘ bigger’ iGPU need higher memory bandwidth . That lends to more memory I/O channels which effectively means bigger package ( if don‘t “ rob Peter to pay Paul” on i/o lanes )
AM5 socket is bigger and iGPU presence should get more uniform ., but still likely will be BGA for packages with largest allocation to iGPU .
Agree, I am pleased as well, as I prefer the optimization for density and lower voltage in the fabrication. But you see my point in that if they did and kept the microarchitecture similar, it would blow everything out of the water. It wouldn't even need to be 20-30W like with Alder Lake's single core at 4.8-5.2GHz. You could even just go to 4-4.2GHz, maybe it would out them at 10-15W with N4X & the associated libraries or something, and killer peak performance at detriment to energy efficiency. Just to show it can be done, as there has been a lot cope about wide architectures and imagined tradeoffs instead of more realistic laziness from AMD/Intel and ISA deficits in area/IPC/power.
This 8Cx Gen 3 will mostly still suck. If they used Arm X2's and N5 they could realize a good 1300-1400 on Geekbench 5 at the 2.7-3.3W range based on MediaTek's Dimensity 9000 results and certain choices they'd make for a computer SoC. Alas, MS didn't have Arm V9 ready for Windows apparently and I think they are saving some of their newer TSMC nodes for phones and next year's Nuvia-based SoC's, which should be on N3 or N4/N5.
Samsung 5NM LPE is garbage and the data consistently show it's inferior to TSMC N7, Arm IP is still solid of course so the X1's and A78's won't be too bad, but yeah, not great on peak performance or efficiency.
i wonder why Apple is titling the next event "Peek Performance" ?
I would call this event this way only if the next M2 would have best core performance or, if Apple would present the next Mac Pro...but that would be an WWDC thing
what do you guys think?!
I would call this event this way only if the next M2 would have best core performance or, if Apple would present the next Mac Pro...but that would be an WWDC thing
what do you guys think?!
I have little doubt that M2 single core will be at least on par with desktop Alder Lake.
Well I think that we will get a "peek" at what they are releasing later. Otherwise it would be "peak". So I am only expecting a peek at the future. It is a word game!
Agree, so you are saying that we can see a peek to the mac pro?
Since the M2 will be release probably this weak, so no
eek into the future: needed
The i9 12900HK scores 1850 points in Geekbench 5 single core. The A15 was 7.2% faster than the A14 in Geekbench 5, so it’s reasonable to think that the M1->M2 will jump from 1720 to 1843 points too. So pretty close (in peak performance only, of course, this is the lowest end Apple Silicon SoC vs the highest end Intel mobile CPU, and they have massively different power consumptions). We’ll see if they can sneak in some extra performance.
Anyway the Peek performance kinda points to Apple presenting the fastest SoC in some category. Maybe the dual M1 Pro/Max? Mac Pro? Maybe just M2 and hinting on how that’ll translate to the future Pro/Max SoCs?