Linus Torvalds drops Intel adopts 32-core AMD Ryzen

If it doesn't relate to Debian, but you still want to share it, please do it here

Linus Torvalds drops Intel adopts 32-core AMD Ryzen

Postby Bulkley » 2020-05-27 17:03

‘My 'allmodconfig' test builds are now three times faster than they used to be’ says Linux overlord

Be sure to watch the video. I priced just one of the components and decided that I can't afford this thing. Still, one can dream.
Bulkley
 
Posts: 5970
Joined: 2006-02-11 18:35

Re: Linus Torvalds drops Intel adopts 32-core AMD Ryzen

Postby Head_on_a_Stick » 2020-05-27 18:47

Ryzen ftw!

Linus knows very well just how crappy Intel's products are, he keeps having to patch them in kernel space...
Black Lives Matter

Debian buster-backports ISO image: for new hardware support
User avatar
Head_on_a_Stick
 
Posts: 12305
Joined: 2014-06-01 17:46
Location: /dev/chair

Re: Linus Torvalds drops Intel adopts 32-core AMD Ryzen

Postby LE_746F6D617A7A69 » 2020-05-27 21:38

Probably he's now very angry, because just after he have switched to 3970x, AMD have released 3990x with 64 cores/128threads :mrgreen:
Bill Gates: "(...) In my case, I went to the garbage cans at the Computer Science Center and I fished out listings of their operating system."
The_full_story and Nothing_have_changed
LE_746F6D617A7A69
 
Posts: 278
Joined: 2020-05-03 14:16

Re: Linus Torvalds drops Intel adopts 32-core AMD Ryzen

Postby Head_on_a_Stick » 2020-05-28 19:00

And Zen 3 is out later this year. Go Team Red!
Black Lives Matter

Debian buster-backports ISO image: for new hardware support
User avatar
Head_on_a_Stick
 
Posts: 12305
Joined: 2014-06-01 17:46
Location: /dev/chair

Re: Linus Torvalds drops Intel adopts 32-core AMD Ryzen

Postby pendrachken » 2020-05-28 19:41

Head_on_a_Stick wrote:Ryzen ftw!

Linus knows very well just how crappy Intel's products are, he keeps having to patch them in kernel space...



Yeah, because AMD has shit the bed zero times in the last 5 years, and is now the king of single thread performance.... Oh wait. No. There have been just as many patches for AMD errata and quirks.


It's almost like using the proper tool for the job or something. Linus does things that need a ton of cores that he doesn't care about single thread performance - buys CPU designed just for that, news at 11. It's not a bash at Intel if someone decides to use AMD because it fits a specific use case better. Same goes for someone choosing an Intel CPU over an AMD one because they use it for task that the Intel is better suited for.

I mean, if that was the case Linux is shit because Photoshop, Lightroom, and the Adobe editing suite that 90% of the world uses doesn't run worth a shit on it... if you can even get it to run in WINE. No matter that it's one of the best server OS choices and is used the world over.
fortune -o
Your love life will be... interesting.
:twisted: How did it know?

The U.S. uses the metric system too, we have tenths, hundredths and thousandths of inches :-P
pendrachken
 
Posts: 1378
Joined: 2007-03-04 21:10
Location: U.S.A. - WI.

Re: Linus Torvalds drops Intel adopts 32-core AMD Ryzen

Postby LE_746F6D617A7A69 » 2020-05-28 21:13

pendrachken wrote:Yeah, because AMD has shit the bed zero times in the last 5 years, and is now the king of single thread performance.... Oh wait. No.
I see, You're an Intel fan...

The funny thing is, that all that stupid flame wars between fans of AMD and Intel are resulting from a trivial fact:
None of those fans is understanding "how the CPU works", so they are sentenced to believe in the sponsored CPU benchmarks results, which, in general are worth shit.

The fundamental mistake made, is an assumption that the same binary executable should work equivalently well on all compatible CPUs.
This is simply not true, because the only property shared between different CPU families and brands is the ISA (Instruction Set Architecture). This means, that all of these CPUs are understanding the same instructions, but they are executing them in a completely different way, because of:
- different instruction pipelines,
- different caching algorithms,
- different prefetching methods,
- different inter-CPU communications methods,
- etc, etc, etc ......

Basically, this means that all the benchmark results You've ever seen in Your life are worth shit.

True comparison of CPUs performance is possible only if You optimize the benchmark algorithm for each particular CPU model and then compare the results.

AthlonXP had "better" performance than Pentium4, because none of the benchmarks was optimized for P4
Intel "Core" series have better IPC, because none of the benchkmarks was optimized for AMD since that time.

So which CPU is better?

I'm using both Intel and AMD CPUs, and I have few projects that are computationally "heavy" - the core of those projects has to be compiled specifically for AMD or Intel to squeeze maximum performance from both CPU models.

For optimized algorithms, the IPC is practically the same for both AMD and Intel.

EDIT: If some benchmark program runs with the same speed on both AMD and Intel CPUs, then it means that such benchmark works equivalently bad in both cases, i.e. it causes the same level of performance loss, but resulting from different reasons.
/EDIT

Regards.

...and sorry for the lenghty post ;)
Bill Gates: "(...) In my case, I went to the garbage cans at the Computer Science Center and I fished out listings of their operating system."
The_full_story and Nothing_have_changed
LE_746F6D617A7A69
 
Posts: 278
Joined: 2020-05-03 14:16

Re: Linus Torvalds drops Intel adopts 32-core AMD Ryzen

Postby Head_on_a_Stick » 2020-05-29 15:39

pendrachken wrote:There have been just as many patches for AMD errata and quirks.

Code: Select all
empty@E485:/sys/devices/system/cpu/vulnerabilities $ grep -R .
spectre_v2:Mitigation: Full AMD retpoline, IBPB: conditional, STIBP: disabled, RSB filling
itlb_multihit:Not affected
mds:Not affected
l1tf:Not affected
spec_store_bypass:Mitigation: Speculative Store Bypass disabled via prctl and seccomp
tsx_async_abort:Not affected
spectre_v1:Mitigation: usercopy/swapgs barriers and __user pointer sanitization
meltdown:Not affected
empty@E485:/sys/devices/system/cpu/vulnerabilities $

I've shown you mine, can I see yours? :P
Black Lives Matter

Debian buster-backports ISO image: for new hardware support
User avatar
Head_on_a_Stick
 
Posts: 12305
Joined: 2014-06-01 17:46
Location: /dev/chair


Return to Offtopic

Who is online

Users browsing this forum: No registered users and 10 guests

fashionable