In 2021, AMD CEO Lisa Su was quoted:
Or maybe RISC-V. It seems that AMD, like Nvdia, could use RISC-V in its GPUs.
AMD Job Posting Hints at Embedded RISC-V CPUs
There is a RISC-V CPU development team at AMD.
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.
[CPU only] Apple M1/2(Max/Ultra) (TSMC 5nm) vs AMD Zen 4 (TSMC 5nm) - Technical Analysis
- Thread starter exoticSpice
- Start date
- Sort by reaction score
In 2021, AMD CEO Lisa Su was quoted:
Tom's Hardware has written a post about the latest Ryzen CPUs with new information from AMD
www.tomshardware.com
AMD Shares New Second-Gen 3D V-Cache Chiplet Details, up to 2.5 TB/s
New images of the I/O Die, too.
The first benchmarks of the 7040 series have finally been appeared.
GB6: 2429/11268
GB6: 2429/11268
LENOVO 21JVESIT10 - Geekbench
Benchmark results for a LENOVO 21JVESIT10 with an AMD Ryzen 7 PRO 7840U processor.
browser.geekbench.com
Seriously impressive results in a low-powered chip. I'm looking forward to full laptop reviews. Would be interesting to see one of these in a Razor laptop.
I think that base was 3,300 Mhz and peak was 5,100 Mhz so the turbo speed might be a lot higher than the rating as is typical for measuring chip power consumption. I typically run my systems at 10% - 30% CPU usage the vast majority of the time so this could mean much better battery life for casual use. I'd guess that you'd have fans spinning loudly if you really cranked it up though.
AMD rates this chip with a TDP of 15-28W and promotes it as Ryzen Z1 Extreme for handheld consoles.
Probably 40-50 watts. But Zen4 has very good efficiency and performs admirably at low power draw. I’d expect these U-series to perform similar to M1 at comparable power draw, mostly because AMD can afford a more advantageous clocks due to their number of cores advantage.
When you make a fair comparison – an M1 Max at N5 (the same process), the 7000 looks somewhere between pretty good and OK. The Max has a slightly lower SC score (around 3% lower, at about 35% lower clock, since we can prettymuch count on the 7000 running at Turbo for SC) and a slightly higher MC score (around 7% higher, running 10 threads on 2E/8P, vs the Ryzen running 16 threads on 8 cores).
This is as close as we have gotten so far to being able to compare high-performance ARM to x86-64 on the same fab process. AMD seems to have done a darn fine job with it, but I think Apple's processors still have a lead here.
This is as close as we have gotten so far to being able to compare high-performance ARM to x86-64 on the same fab process. AMD seems to have done a darn fine job with it, but I think Apple's processors still have a lead here.
I heard that Intel's earnings were terrible this week even though I thought that they were winning back marketshare with discounting.
I like the idea of laptops that run cool with a light to medium workload as most people don't require high performance from laptops and these new AMD chips look pretty attractive for Windows laptops.
The comparison of M2 and 7840U results is fascinating. M2 scores some big wins (Navigation, Clang, Object Remover, Photo Filter and Structure from Motion), but so does 7840U (Photo Library, Object Detection and Ray Tracer). The strangest result is Photo Filter because 7840U wins on multicore by a wider margin than other benchmarks, yet loses by a lot on single core.
I can understand 7840U winning when the benchmark uses SIMD instructions, but what explains M2 winning?
I can understand 7840U winning when the benchmark uses SIMD instructions, but what explains M2 winning?
M2 is a faster CPU, 7840U has more cores and can operate them at higher clock for brief periods of time.
That doesn't explain the strange results. For example, M2 loses in the single-core Object Detection test, but wins with multiple cores.
It appears that M2 results are independent of the amount of RAM.
One of the big wins of M2 is Navigation. According to the documentation, the Navigation test uses Dijkstra's algorithm.
Geekbench 6 CPU Workloads https://www.geekbench.com/doc/geekbench6-cpu-workloads.pdf
What makes M2 so good in the Dijkstra's algorithm?
Such a job would involve a whole lot of conditional branching, so perhaps Apple/ARM's branch prediction model is that much better, and branch prediction misses might also incur less of a penalty.
Those are high power parts; I doubt very much that these low power parts are affected by this problem and I've read that the actual numbers are pretty small. I love Gamers Nexus for breaking this stuff and then trying it out themselves.
No, there are some non-X3D parts that have the problem too. I got this from Paul's Hardware this morning.
Golden Pig Upgrade has shared a review of the 7840HS, a chip that AMD rates with a TDP of 35-45W. Unfortunately, he used benchmarks useful for gaming.
www.bilibili.com
R7 7840HS深度评测：Zen4大提升，但核显还是叫RDNA2.5吧_哔哩哔哩_bilibili
一路磕磕绊绊总算测完了，主包这周又是肠胃炎又是嗓子哑实在是太累，估计后面会歇一段时间了。, 视频播放量 283470、弹幕量 417、点赞数 5280、投硬币枚数 1990、收藏人数 1142、转发人数 766, 视频作者 金猪升级包, 作者简介 既不独立，也不客观，更不公正，相关视频：锐龙7000深度评测：强，但还不够强...，Zen4上笔记本了，AMD R9-7845HX / R9-7945HX 移动处理器性能分析，大家久等了R7 7840HS即将发布！丰富接口；优异能耗；超强核显~，AMD Ryzen 7 7840HS 游戏基准测试 [已测试 21 款游戏]，锐龙 R7...
Bummer. Is there more info available yet as to what processors / which circumstances lead to mentioned meltdowns?
Both shows indicated that this was not common. AMD has issued updates to the motherboard companies so they should be putting out updates to prevent this from happening in the future. Not really helpful for those that had this though.