sliken

13 ms·

211.
by sliken 2y ago
Apple silicon manages around 50% (giver or take) for the CPUs.
212.
by sliken 2y ago
Didn't find anything clearly stating one way or another, but the CCD is the same between ryzen and epyc, so there's certainly the possibility. I dug around a bit, and it seems Ryzen doesn't get it. I guess that makes sense,
213.
by sliken 2y ago
Dimms are dumb. Not sure, but maybe rambus helped improve this. Dimms are synchronous and each memory channel can have a single request pending. So upon a cache miss on the last level cache (usually L3) you send a row, column, wait 60ns
214.
by sliken 2y ago
I've been paying close attention, found various hints at anandtech (RIP), chips and cheese, and STH. It doesn't make much difference to most apps, but I believe the single CCD (like the 9700x) has better bandwidth to IOD then thei
215.
by sliken 2y ago
I do wish I could pay $25 a month for my web content to be ad free. Portioned out to websites I actually spent time reading.
216.
by sliken 2y ago
My understanding is the single CCD chips (like the 9800x3d) have 2 IF links, while the dual CCD chips (like the 9950x) have 1. Keep in mind these CCDs are shared with turin (12 channel), threadripper pro (8 channel), siena (6 channel), thr
217.
by sliken 2y ago
Well there's much more to memory performance than bandwidth. Generally applications are relatively cache friendly, thus the X3D helps a fair bit, especially with more intensive games (ones that barely hit 60 fps, not the silly game be
218.
by sliken 2y ago
Or maybe you should follow the recommendations of various government agencies (including the FBI) and install an ad blocker.
219.
by sliken 2y ago
In theory, yes. But in the real world the bottleneck of the same 128 bit wide memory, interface that's been popular way back since the time of dual core chips. Less cache misses (on popular workloads) helps decrease power and increase
220.
by sliken 2y ago
That was the result of an previous thread on asahi. It was the only time I've been embarrassed by the HN community and it's treatment of people contributing to open source.
221.
by sliken 2y ago
It does seem clear that financial issues are a major contributor. Parents have been getting older before having children, presumably because it's taking longer to have financial stability. I know it was a factor with our kid. If you
222.
by sliken 2y ago
Agreed. I was sorely tempted by the Mac studio, but ended up with a 96GB ram Ryzen 7900 (12 core) + Radeon 7800 XT (16GB vram). It was a fraction of the price and easy to add storage. The Mac M2 studio was tempting, but wasn't refres
223.
by sliken 2y ago
The cost is a rate, like $2 per hour, not a purchase price. So faster CPUs get work done more quickly and may justify a higher cost per hour. Ah, the graph is wacky, but the text makes sense, looks like a disconnect: 1. C4A Axion: $2.1
224.
by sliken 2y ago
Sadly these days it's a waste to upgrade a single generation. There's very few workloads that work poorly on gen x-1 and work well on gen x. So while frustrating to people that want to geek out on the tiny details, the real marke
225.
by sliken 2y ago
Ugh, should have caught the bit vs byte, thanks. The M4 max is definitely not a 4090 killer, does not match it in any way. It can however work on larger models than the 4090 and have a battery that can last all day. My memory is a bit fuzz
226.
by sliken 2y ago
Indeed. RIP Anandtech. I've seen bandwidth tests since then that showed similar for newer generations, but not the m4. Not sure if the common LLM tools on mac can use CPU (vector instructions), AMX, and Neural engine in parallel to
227.
by sliken 2y ago
Heh, ok, maybe slightly different. But apple spec claims 546GB/sec which works out to 512 bits (64 bytes) * 8533. I didn't think the point was 8533 vs 8800. I believe I saw somewhere that the actual chips used are LPDDR5X-8533.
228.
by sliken 2y ago
At a linux conference I saw many macbooks. Talked to a few, they just ran linux in a VM full screen for programming and related. Then used OSX for everything else (office, outlook, teams, work enforced apps, etc). They seemed very happy
229.
by sliken 2y ago
Yes, you could buy a brand new (announced weeks ago) AMD Turin. 12 channels of DDR5-6000, $11,048 and 320 watts (for the CPU) and get 576GB/sec peak. Or you could buy a M3 max laptop for $4k, get 10+ hour battery life, have it fit in
230.
by sliken 2y ago
> Memory interface width of modern CPUs is 64-bit (DDR4) and 32+32 (DDR5). Sure, per channel. PCs have 2x64 bit or 4x32 bit memory channels. Not sure I get your point, yes PCs have 64 bit cache lines and apple uses 128. I wouldn'
231.
by sliken 2y ago
No.
232.
by sliken 2y ago
M2 pro has 256 bit wide memory, mostly benefiting the GPU perf. M3 pro has 192 bit wide memory, GPU improvements mostly offset the decrease in memory bandwidth. This leads to memory options like 96GB. M4 pro has 256 bit wide memory, thus t
233.
by sliken 2y ago
Yes, upgrading from a m3 max to a m4 max would be a waste.
234.
by sliken 2y ago
I think you are confusing channels and dimms. The vast majority of any x86 laptop or desktops are 128 bits wide. Often 2x64 bit channels up till last year or so, now 4x32 bit DDR5 in the last year or so. There are some benefits to 4 chann
235.
by sliken 2y ago
M4 max in the MBP (today) and in the Studio at some later date.
236.
by sliken 2y ago
Fewer libraries? Any that a normal LLM user would care about? Pytorch, ollama, and others seem to have the normal use cases covered. Whenever I hear about a new LLM seems like the next post is some mac user reporting the token/sec.
237.
by sliken 2y ago
The new idea is having 512 bit wide memory instead of PC limitation of 128 bit wide. Normal CPU cores running normal codes are not particularly bandwidth limited. However APUs/iGPUs are severely bandwidth limited, thus the huge numbe
238.
by sliken 2y ago
Sort of, obviously quite a few games are optimized for the PS5 and Xbox series X. GPU cores are generally identical between the iGPUs and the discrete GPUs. Adding a PCIe bus (high latency and low bandwidth) and having a separate memory po
239.
by sliken 2y ago
Dunno, seems the opposite. Dyson are plastic, don't last long, and generally aren't serviceable. Apples have been improving repairability, their laptops have many replaceable parts, and generally last longer than PCs. Compare new
240.
by sliken 2y ago
Well the M4 pro crushes the M3 pro. Presumably the M4 max will be a better than the M3 max. No idea if the Ultra will come out, apple hasn't done that since the M2, even if they do it's likely to be 6 months away.

More ›