With the release of Alder Lake less than a week away and the "Lakes" thread having turned into a nightmare to navigate I thought it might be a good time to start a discussion thread solely for Alder Lake.
I wonder if that's because AMD sees the APUs as their target for the low-end. Honestly, I don't think that's a bad approach if (a) they'd update their APUs faster, and (b) not saddle them with older technology so often. I'd love if either Intel or AMD would give a CPU with a built-in GPU that could just do 1080p medium in your average AAA game. It's why I'm a bit anxious to see how the 6000 series APUs from AMD perform!
APU with V-cache (even just 32MB) would get there faster.I do think with RDNA2 and DDR5 1080p medium on less stressful AAA games may become more of a possibility.
While I don't game, I use enough graphics (such as 3D models) that I too would like CPUs with just a bit more graphical capabilities. For the most part you can get a desktop chip with a great CPU but crappy integrated graphics or a mobile chip with a barely OK CPU and great integrated graphics. There isn't much in between. Heck the cheapest mobile Celeron Alder Lake has more iGPU execution units than the top desktop Alder Lake (48 vs 32, although yes the mobile Celeron has them clocked lower).I'd love if either Intel or AMD would give a CPU with a built-in GPU that could just do 1080p medium in your average AAA game.
While I don't game, I use enough graphics (such as 3D models) that I too would like CPUs with just a bit more graphical capabilities. For the most part you can get a desktop chip with a great CPU but crappy integrated graphics or a mobile chip with a barely OK CPU and great integrated graphics. There isn't much in between. Heck the cheapest mobile Celeron Alder Lake has more iGPU execution units than the top desktop Alder Lake (48 vs 32, although yes the mobile Celeron has them clocked lower).
I'm not a fan of adding a video card for hundreds of dollars (now thousands) and that uses dozens to hundreds of Watts of power for the limited 3D uses that I do have.
I'd gladly drop 2 P cores in Alder Lake desktop chip for a doubling of the iGPU execution units. The die size is roughly on par. The 12700H is the closest to what I'd like to see. But that is still not really the best solution for a desktop computer that has so much more power available to give.
We actually can't say that AMD is behind in the sense that their current product is lacking. We're still waiting on zen4 and then we'll see how much AMD is behind or on par. Intel released Rocket Lake, Tiger Lake and Alder Lake in very short spans while AMD is still on zen3 products. As far as I can see Intel has gained a proper performance gain from AMD in Alder Lake whereas they were almost neck to neck with zen3 with Rocket lake even if if was a power hog. And as a way to get more MT throughput they exchanged the 10 core part with 8 GC + 8 GM cores.Intel got ahead in performance, as expected.
But how far behind is AMD, actually? Would allowing the chips to use as much energy as possible like Intel does get it half the way to parity again? Or is the bottleneck the architecture?
P.S: I think adding more power to zen 3 won't do anything.
Definitely it will help a little, but mobile ryzen chips are geared towards power efficiency for longer battery life.Not on desktop, but at least on mobile couldn't help a bit?
So amazing to see M1 Pro's arse getting whooped in almost everything. The only places where it wins is Export PDF to PNG and Premiere (good showing at only 60W). M1 Pro is still the battery king but I would rather have higher performance.
Further refining the HandBrake engine to support native 10 and 12-bit encodes, including HDR10 metadata passthru.
[*]Improvements to hardware encoding functionality for Intel QuickSync, AMD VCN and Qualcomm ARM devices. (Thanks to these companies all for supporting the development in HandBrake!)
[*]Adds support for Apple Silicon based macs.
[*]Adds support for Qualcomm ARM64 devices running Windows (HandBrakeCLI only for now. Windows UI is coming later!)
[*]Improvements to subtitle handling.
[*]UI/UX improvements for all 3 platforms.
[*]As usual, hundreds of other changes and tweaks to the app. See the full release notes below for details!
For the Mac Pro, Apple needs SMT or 20 cores minimum to declare themselves triumphant over Alder Lake.
Here's another one, this time with the 12900HK.
At least we can put to rest the notion that the M1 Pro/M1 Max is "fastest at everything". From those numbers, it looks like the M1 Pro and Max would lose Handbrake to previous-gen processors.
Another thing to consider is that even at 50W max, M1 Pro is hitting high 90s in temperature. There is virtually no thermal headroom left for Apple to add more cores since their highly dense SoC is producing too much heat to dissipate quickly. For higher core count, they will need to go with water cooling or die-stacking to create more space between the different hot spots or just wait till 3nm.Let's just be fair here: the M1 Pro struggles to pull more than 50W. During R23 using wall power, the 12900HK pulls 141W. It isn't clear from the video you linked what was the power delta between systems in each test, but that should give us some idea of what's really going on here. The 12900HK is a hungry beast.
At least we can put to rest the notion that the M1 Pro/M1 Max is "fastest at everything". From those numbers, it looks like the M1 Pro and Max would lose Handbrake to previous-gen processors.
The M1 Pro draws more like 30-35W in cinebench MT (as reported by some reliable macrumors posters). Anandtech reports 34W package power (ram included I suppose). 40W is AC active power. (And not to mention that cinebench isn't well optimised for ARM. It uses a translation layer for intel's embree ray tracing API).From HWUB's testing, Alder Lake-H + TGL-H seem to scale much better with power than Cezanne does, unless TSMC 6nm is a radically different process I don't see how that changes with Rembrandt.
The Macbook Pro fans run a moderate speed on a design that is quite thin. 90°C is in the range of normal operation for a CPU. Apple is fine with letting the CPU hit the high 90°C.Another thing to consider is that even at 50W max, M1 Pro is hitting high 90s in temperature. There is virtually no thermal headroom left for Apple to add more cores since their highly dense SoC is producing too much heat to dissipate quickly. For higher core count, they will need to go with water cooling or die-stacking to create more space between the different hot spots or just wait till 3nm.
Adding more cores would make the current fans inadequate for cooling. They might have to go with watercooling or huge fans (like in PSUs) or air blower-type solution that sounds like a jet engine.The Macbook Pro fans run a moderate speed on a design that is quite thin. 90°C is in the range of normal operation for a CPU. Apple is fine with letting the CPU hit the high 90°C.
They have some headroom. Adding more cores won't radically increase the power density for cooling.
The 12900HK is a hungry beast.
From those numbers, it looks like the M1 Pro and Max would lose Handbrake to previous-gen processors.
We're looking at the M1 Pro in a 0.66 in. (1.68 cm) thick 4.7 lb. (2.1 kg) laptop with a cooling system designed to not exceed 37.4 dBA and a skin temperature of 41 ºC at maximum load under normal conditions. Do you really think there's no thermal headroom left for this architecture based on that?Another thing to consider is that even at 50W max, M1 Pro is hitting high 90s in temperature. There is virtually no thermal headroom left for Apple to add more cores since their highly dense SoC is producing too much heat to dissipate quickly. For higher core count, they will need to go with water cooling or die-stacking to create more space between the different hot spots or just wait till 3nm.
A bigger heatsink and a louder fan would decrease the temperature down to what, 10 degrees less?We're looking at the M1 Pro in a 0.66 in. (1.68 cm) thick 4.7 lb. (2.1 kg) laptop with a cooling system designed to not exceed 37.4 dBA and a skin temperature of 41 ºC at maximum load under normal conditions. Do you really think there's no thermal headroom left for this architecture based on that?
The cooling system doesn't need to be any louder, and it doesn't need to reduce the maximum temperature at all—it just needs to be able to dissipate more heat energy in a given amount of time. We have no idea how much power you can dump into the M1 Pro / Max before thermal density becomes the primary issue.A bigger heatsink and a louder fan would decrease the temperature down to what, 10 degrees less?
I don't see how Apple's core philosophy is a liability. There's nothing stopping them from upping the clock speeds significantly, without entering the GHz race mind you. The thing's you talk about in the following paragraph...But for stuff like gaming or heavy throughput workloads like software encoding, a super wide core with comparatively much lower clock speeds is now a liability. It goes in the other direction as well too, because those same attributes are what makes the M1 so strong in compilation.
...are things anybody would like to have in a processor. The reason why it is not as good as x86 in Handbrake encoding is because there is no hand-tuned assembly code for it, just basic support.I'm just a layman, but I think the reason why the M1 is so good at compiling is because it has very high IPC, very accurate branch predictor, relatively short pipeline (so smaller penalties for misses) and humongous amounts of cache.
True. I was merely wondering how Apple will add additional cores should they need them to compete with Intel/AMD in future. Or they might refuse to compete on core count and just say they have the best battery life in the industry. That would still work great for them.If you want that kind of a "laptop" you will never be able to buy it from Apple, so it is pointless to speculate about whether the Macbook's cooling system could be adapted to become one.