5 ms·
Few things. 1. M1 in Mac Mini is estimated to be 20W to 25W. 2. M1 is 4 Performance Core + 4 Efficient Core. Compared to 8 Performance Core in Ryzen 3. Ryzen
by ksec 6y ago
Few things.
1. M1 in Mac Mini is estimated to be 20W to 25W.
2. M1 is 4 Performance Core + 4 Efficient Core. Compared to 8 Performance Core in Ryzen
3. Ryzen is 8 Core 16 Thread. M1 is 8 Thread.
4. Worth mentioning again, M1 is on 5nm, Ryzen is on 7nm.
- alpanka 6y agoNote that AMD has been beating Apple on single core performance since day 1. See the comparison between 4700u, M1 and micromagic Risc-v: https://arstechnica.com/gadgets/2020/12/new-risc-v-cpu-claims-recordbreaking-performance-per-watt/ https://arstechnica.com/gadgets/2020/12/new-risc-v-cpu-claim... I assume the efficient cores on M1 gives it an edge on power usage.
- stefan_ 6y agoPower usage follows directly from the 5nm process.
- alpanka 6y agoIt's a factor, but not the only one. It also depends on the foundary and the exact process and of course the CPU design.
- agumonkey 6y agoIn which case will make a very nice marketing speech for AMD
- osti 6y agoSo you decide to cherry pick one single thread benchmark out of many to decide that AMD wins in single thread lol. In most benchmark, m1 is much faster than 4800u in ST.
- api 6y agoApple just entered an extremely competitive market with the M1, which is good because we win if this market is more competitive.
- ben-schaaf 6y agoThey didn't enter any market they weren't already in. They've been selling laptops and sff desktops for years and they certainly haven't and aren't going to started selling CPUs. Not to say that the competition isn't good, they just haven't entered into any new markets.
- BugsJustFindMe 6y ago> Note that AMD has been beating Apple on single core performance since day 1 On EEMBC CoreMark. But CoreMark is not a meaningful benchmark outside of embedded devices. From your linked article: "CoreMark focuses solely on the core pipeline functions of a CPU, including basic read/write, integer, and control operations. This specifically avoids most effects of system differences in memory, I/O, and so forth." Memory & I/O are integral to CPU design, which is why AMD's own marketing material emphasizes it so much.
- nojito 6y agoThe power comparison isn't Apples to Apples. https://www.anandtech.com/show/16252/mac-mini-apple-m1-tested https://www.anandtech.com/show/16252/mac-mini-apple-m1-teste...
- PragmaticPulp 6y agoAs an engineer, I love examining the details. I think the M1 is a great processor, but the performance advantages have been greatly exaggerated. As an end-user, I just don't care if my CPU is 4 threads or 16 threads or 5nm or 7nm or 15W or 20W. I only care about the price, how fast it handles my workloads, and how long the battery lasts. I think AMD understands this angle well.
- mhh__ 6y agoI think some of the exaggeration has come from people who only use Apple products basically seeing what the cutting edge actually is, e.g. We've historically been the same performance or worse, now our tribe is the fastest - spurred on by much faster benchmarks. Anyway I look forward to websites slowing down as M1 becomes more popular amongst Devs who shouldn't need anything that fast in the first place (ideally)
- ece 6y agoThat doesn't sound so good, but I can do one better. The 4 FP EUs/core on the M1 (x64 has 2) that likely lead to this result might also lead other languages to use more floats and doubles.
- mhh__ 6y agoI thought M1s SIMD throughput was lower?
- ece 6y agoIt is, but the Firestorm cores can do 2x more scalar FP executions per clock than Zen 3. Still, for a full browser benchmark like Speedometer 2, I would've liked to see better results from 5800H/U given their large cache bumps from last gen. https://www.anandtech.com/show/16226/apple-silicon-m1-a14-deep-dive/2 https://www.anandtech.com/show/16226/apple-silicon-m1-a14-de... https://www.anandtech.com/show/16214/amd-zen-3-ryzen-deep-dive-review-5950x-5900x-5800x-and-5700x-tested/4 https://www.anandtech.com/show/16214/amd-zen-3-ryzen-deep-di...