The money is in the server market. AMD currently has a very compelling offering there, but many companies will be very reluctant to switch vendors.
It's that same "Noone gets fired for buying Intel" mediocre mentality...
The money is in the server market. AMD currently has a very compelling offering there, but many companies will be very reluctant to switch vendors.
Yes, but Apple uses Intel chips in the Macs right now. That is about 10% of the total PC market. That will be a blow to Intel's market share and revenue.
Hahahha.
Of course, there shouldn't be absolutely any deny about AMDs leadership when it comes to productivity/rendering performance.
Still,
I long hesitated between AMD and Intel platform for my early 2020 build.
Since my main computer is 80% about gaming, I ended up with a Z390 / 9700K / 32Go 3600Mhz set up. Why? Because Intel CPUs are still ahead when it comes to FPS/pure gaming performance (and only this). Pretty obvious when you look at 1080p benchs. Then, the gap reduces when you go up to 1440p/UHD
For my future build, I could turn around for some X670/AMD 4000 build. But I will only if Gaming performance benchmarked (same as "measured") comes closer to what Intel 10th Gen has to offer.
By the way, the aforementioned build is a very tiny SFF/itx one. Here it is:
Yes, that is what I was saying earlier.Macs are on their way for the ARM... At the end of the year Intel will lost this market.
This isn't the first time AMD has taken Intel by surprise. Let's not forget it was AMD that brought us cheap 64-bit computing. Even when not in the lead, AMD has always been there alongside.
View attachment 67121
What was that about computers getting smaller?
As a rough estimate, we can suppose the 8088 has a throughput of 1M instructions per second and the EPYC 1 instruction per cycle per core at 3 GHz. A 64-core EPYC (the pictured one is a 16-core) then has about 200,000 times higher instruction throughput. Since each instruction can also accomplish a lot more, it is probably safe to say the EPYC is at least 1 million times faster. The 8088 DIP package measures 53x16 mm. A million of these would have an area of 848 m². Since a little space between the chips is necessary, we can round this up to 1000 m². That's roughly the size of 5 tennis courts. Powering this monstrosity would require about 2 MW while the top EPYC needs 280 W.I wonder what the foot print for the D8088 would be if you had to put enough of them together to equal the compute power of that single Epyc?
My CPU (i9-9900K) is actually an important part of my music playback.
I use HQPlayer (on the fly convolution for digital room correction plus upsampling to DSD) and some of their newer DSD modulators need running speed (max boost) close to 5GHz for 2 cores. I don't overclock.
I haven't seen any AMD's with running speed at 5GHz without overclocking?
Are we talking about AMD catching Intel by surprise in mobile CPU's? Or everything?
I would imagine for peak gaming performance, the i9‑10900K is king? Especially when paired with top Nvidia RTX GPU?
You can't do that with Intel!
Intel absolutely dominates single threaded performance. I am running 3700x and my Jriver with digital EQ (FIR at 262k taps - eat that Chord!) is able to barely play real time without dropouts. Single thread performance of 3700x is good enough for now and when 4900x comes around I will pop that in and get an instant boost. You can't do that with Intel!
Intel absolutely dominates single threaded performance. I am running 3700x and my Jriver with digital EQ (FIR at 262k taps - eat that Chord!) is able to barely play real time without dropouts. Single thread performance of 3700x is good enough for now and when 4900x comes around I will pop that in and get an instant boost. You can't do that with Intel!
Hi, do you mean you can't do that same DSP with an i9-9900K (or faster)?
I believe he is referring to Intel’s constantly changing motherboard chipset and sockets making newer generation chips incompatible. The new 10th generation pentium chips are nothing more than a mild update to the 9th generation, but require a new bios chipset and in turn motherboard. Conversely, AMD has pledged and followed through that their motherboards will be compatible for several generations of cpus. I believe the current 3rd gen Ryzen cpus will operate on first gen boards and AMD says 4th gen will also.
How can someone allergic to BS support Intel? The company that still doesn't use ECC for their caches or support ECC RAM on their non Xeon lines, still uses subpar thermal paste under their IHS instead of solder, cheats in benchmarks, does shady deals with OEMs to keep their monopoly and use very bad libel disguised as marketing (e.g. "lol, AMD is gluing chip together. Nevermind that we did too after a few months").IIRC, when AMD first started challenging Intel in the CPU market they shipped a lot of bad CPUs that had be replaced. It allowed them to claim the numbers being shipped against Intel. Along with many years of inferior measured performance compared to Intel has always made me want to stay away from AMD. I later gave them some preference for video cards that eventually went away in favor of Nvidia.
I've built and rebuilt a lot of computers over the years and only one had an AMD CPU from scratch. My dad used it for a decade for internet and email. The only reason it was replaced was because I didn't want to deal with him using Win 7 after its retirement and we decided it was time for a new computer. It was a decent machine for that sort of thing but it had weird quirks and failures. If I wanted something for power and dependability it always had an Intel CPU. Including the times I've purchased assembled desktops and laptops. I realize times have changed for both companies and I'm biased, but until I have a very good reason to switch I'll go Intel for a serious build. Partly because AMD started with BS claims of success based on shipping a bunch of faulty CPUs.
It doesn't though. Here's a non synthetic open-source single-threaded benchmark that souldn't be only about SIMD unit performance:Intel absolutely dominates single threaded performance. I am running 3700x and my Jriver with digital EQ (FIR at 262k taps - eat that Chord!) is able to barely play real time without dropouts. Single thread performance of 3700x is good enough for now and when 4900x comes around I will pop that in and get an instant boost. You can't do that with Intel!
How can someone allergic to BS support Intel? The company that still doesn't use ECC for their caches or support ECC RAM on their non Xeon lines, still uses subpar thermal paste under their IHS instead of solder, cheats in benchmarks, does shady deals with OEMs to keep their monopoly and use very bad libel disguised as marketing (e.g. "lol, AMD is gluing chip together. Nevermind that we did too after a few months").
And best of all, pointing to AMD's shipping of faulty CPUs (that they did, and it affected Gentoo users like me the most, but we got free RMAs) but conveniently ignoring the endless stream of vulnerabilities that Intel's shoddy chips continue to exhibit. We even got another this morning: https://www.guru3d.com/news-story/c...rmation-to-be-extracted-from-other-cores.html
Intel is actually used by the audiophiles of the computer world, the ones who don't really know how computers work but "they know their ears and brands". AMD (even Piledriver) is used a lot by libre software/OS devs who actually know what it's all about:
for example, Linus Torvalds (https://lore.kernel.org/lkml/CAHk-=whan1CiRtcgBt-5SkW-ga_GeLH5+AO26RmK7vOA5yw9ng@mail.gmail.com/T/#u) and Matthew Dillon (http://lists.dragonflybsd.org/pipermail/users/2018-August/357852.html).
It doesn't though. Here's a non synthetic open-source single-threaded benchmark that souldn't be only about SIMD unit performance:
https://www.guru3d.com/articles-pages/intel-core-i9-10900k-processor-review,14.html
It doesn't really show a "domination". And most "real-world" software like browsers (or anything doing networking) will be a lot worse due to the syscall overhead of Intel's combined vulnerabilities.
Really, you should consider that Windows has a poor scheduler that doesn't deal well with the "internal NUMA" setup of Ryzen, that benchmarks could be (are often) compiled using ICC with Intel's math libraries and other stuff that doesn't allow you to really gauge hardware performance.
Now, I may come off as a fanboy, but I just really know about this and don't want people to be the audiophiles/living jokes of other crowds while they laugh at the original ones.
Intel only has the edge in two domains, right now: SIMD performance and (slightly) clock speeds. Still, you'd be buying from crooks selling half backed designs with more holes than Swiss cheese.
Curiously, this was the same argument deployed by NVIDIA against AMD in the GPU battle. Of course they actually brought the receipts and micro-stutter turned out to be a real thing that eventually expanded the range of measurements that serious reviewers use to evaluate new hardware. I'm not holding out hope for Intel to back up the talk.While there might be faster CPU's, you cannot deny the elan and smoothness one feels in using a proven Intel based machine. There is more to computing than specs.
Hmmmmm, NO, I don't think that is going to fly.
How can someone allergic to BS support Intel? The company that still doesn't use ECC for their caches or support ECC RAM on their non Xeon lines, still uses subpar thermal paste under their IHS instead of solder, cheats in benchmarks, does shady deals with OEMs to keep their monopoly and use very bad libel disguised as marketing (e.g. "lol, AMD is gluing chip together. Nevermind that we did too after a few months").
And best of all, pointing to AMD's shipping of faulty CPUs (that they did, and it affected Gentoo users like me the most, but we got free RMAs) but conveniently ignoring the endless stream of vulnerabilities that Intel's shoddy chips continue to exhibit. We even got another this morning: https://www.guru3d.com/news-story/c...rmation-to-be-extracted-from-other-cores.html
Intel is actually used by the audiophiles of the computer world, the ones who don't really know how computers work but "they know their ears and brands". AMD (even Piledriver) is used a lot by libre software/OS devs who actually know what it's all about:
for example, Linus Torvalds (https://lore.kernel.org/lkml/CAHk-=whan1CiRtcgBt-5SkW-ga_GeLH5+AO26RmK7vOA5yw9ng@mail.gmail.com/T/#u) and Matthew Dillon (http://lists.dragonflybsd.org/pipermail/users/2018-August/357852.html).
It doesn't though. Here's a non synthetic open-source single-threaded benchmark that souldn't be only about SIMD unit performance:
https://www.guru3d.com/articles-pages/intel-core-i9-10900k-processor-review,14.html
It doesn't really show a "domination". And most "real-world" software like browsers (or anything doing networking) will be a lot worse due to the syscall overhead of Intel's combined vulnerabilities.
Really, you should consider that Windows has a poor scheduler that doesn't deal well with the "internal NUMA" setup of Ryzen, that benchmarks could be (are often) compiled using ICC with Intel's math libraries and other stuff that doesn't allow you to really gauge hardware performance.
Now, I may come off as a fanboy, but I just really know about this and don't want people to be the audiophiles/living jokes of other crowds while they laugh at the original ones.
Intel only has the edge in two domains, right now: SIMD performance and (slightly) clock speeds. Still, you'd be buying from crooks selling half backed designs with more holes than Swiss cheese.
How can someone allergic to BS support Intel? The company that still doesn't use ECC for their caches or support ECC RAM on their non Xeon lines, still uses subpar thermal paste under their IHS instead of solder, cheats in benchmarks, does shady deals with OEMs to keep their monopoly and use very bad libel disguised as marketing (e.g. "lol, AMD is gluing chip together. Nevermind that we did too after a few months").
And best of all, pointing to AMD's shipping of faulty CPUs (that they did, and it affected Gentoo users like me the most, but we got free RMAs) but conveniently ignoring the endless stream of vulnerabilities that Intel's shoddy chips continue to exhibit. We even got another this morning: https://www.guru3d.com/news-story/c...rmation-to-be-extracted-from-other-cores.html
Intel is actually used by the audiophiles of the computer world, the ones who don't really know how computers work but "they know their ears and brands". AMD (even Piledriver) is used a lot by libre software/OS devs who actually know what it's all about:
for example, Linus Torvalds (https://lore.kernel.org/lkml/CAHk-=whan1CiRtcgBt-5SkW-ga_GeLH5+AO26RmK7vOA5yw9ng@mail.gmail.com/T/#u) and Matthew Dillon (http://lists.dragonflybsd.org/pipermail/users/2018-August/357852.html).
It doesn't though. Here's a non synthetic open-source single-threaded benchmark that souldn't be only about SIMD unit performance:
https://www.guru3d.com/articles-pages/intel-core-i9-10900k-processor-review,14.html
It doesn't really show a "domination". And most "real-world" software like browsers (or anything doing networking) will be a lot worse due to the syscall overhead of Intel's combined vulnerabilities.
Really, you should consider that Windows has a poor scheduler that doesn't deal well with the "internal NUMA" setup of Ryzen, that benchmarks could be (are often) compiled using ICC with Intel's math libraries and other stuff that doesn't allow you to really gauge hardware performance.
Now, I may come off as a fanboy, but I just really know about this and don't want people to be the audiophiles/living jokes of other crowds while they laugh at the original ones.
Intel only has the edge in two domains, right now: SIMD performance and (slightly) clock speeds. Still, you'd be buying from crooks selling half backed designs with more holes than Swiss cheese.