AMD Ryzen 9 5900X + Ryzen 9 5950X Dominate On Linux

[Tomin]

In the previous generation AMD lowered the launch prices because they were in a position where they thought that it would help them sell more CPUs (and thus get more money). Now they upped their launch prices because they think they can get more money that way. They even didn't go insane with the prices but they are still quite well in line with what they offer. Some people will pick competitor's product and some will stick with the older generation and that's totally fine.

Furthermore it's not surprising that previous generation can give more value for the price than the latest. They still want to sell their stock empty. Launch price tends to be higher than what it is once next generation is published and that's just how it works. Also this 50$ more for launch price is quite modest in my opinion. It's very hard to see anything wrong with this.

[f0rmat]

I was thinking the same thing f0rmat. I worked in the Silicon Valley during my career and had been looking for Intel alternatives, as Intel was pretty much universally despised by engineers for their unethical business practices even then. I'd experimented with VIA, Cyrix, and AMD processors, but until the K7 they were all so bad I still had to use Intel. But wow, when the K7 arrived it was like a dream come true. Suddenly the mighty Intel was back on its heals, and actually forced to innovate and compete. And I've used AMD processors exclusively since then.

Of course the unfortunate Bulldozer blunder, and subsequent iterations, were difficult to take for many years. However I still remained loyal to AMD, confident that one day they would rectify those errors.

And today is that day!

Long live AMD! And down with the Intel tyrants!

Cyrix and Via...those are names I have not heard in awhile. I toyed with a Cyrix 6x86 one time, but did not buy it. Having four kids, you picked carefully - the funds were not there for experimentation. My first real build was with an Athlon 700 on a FIC motherboard because that was the only company that built a motherboard at the time for a K7. The forums were alight with theories (which I do not know if they were true) that Intel had put out the word to manufacturers not build motherboards for the Athlon (Asus actually had a prototype that they later put into production). What I do know was true is that AMD had to build its own chipset for it because no one else built them. In some ways, those were more difficult days. You not only had to be concerned about the processor and the motherboard, but you had to worry about the north bridge and the front side bus (remember matching that with the RAM speed). AMD built the 760 series of north bridges - all solid, if a bit conservative. Serverworks finally started building AMD chipsets (another name from the past). IIRC, Via also used to build north bridges for AMD. What is ironic in today's environment is that nVidia built one of the best north bridges for the K7 (and I think K8 series) that was ever put on AMD motherboards. I had a couple of builds with nVidia chipsets including a Tyan board dual athlon server. That first Athlon 700 (as well as the Tyan server) build was finally retired because it was used for so long that I could no longer update the hardware.

I even remember changing EDO RAM in a Gateway 200MHz Pentium MMX computer in my 30s. S**t, I am old.

[torsionbar28]

Ryzen 5 3600 - $200
Ryzen 5 5600X - $300, a whopping 50% price increase (since 5600 has never been announced or hinted at).

Quite a warranted comparison to be fair.

I fail to see the problem. Both products are readily available today. I literally just last week bought a Ryzen 5 3600 brand new for $189 to upgrade my PC with. I thought it was a better value than $299 for the 5600X. But I'm glad the 5600X is available, as Zen3 is a clearly superior architecture and shows the tremendous advancements AMD is continuing to make with Zen. Choice is good. Nobody is forcing you to buy Zen3. If you think Zen2 is better value, as I did, then buy that instead. I suspect there will be Zen2 chips in the channel for a while yet.

Or you can just buy intel if you like paying more to get less. After all, AMD is not competing with itself. How the prices compare to its own past products is irrelevant. How AMD's current prices compare with intel's current prices is the only thing that matters to the market.

[CochainComplex]

I was comparing the prices of the K versions of i5 and i7 cpu's from skylake to comet lake. There was a slight price increase with the coffee lake generations, which is probably because of the additional cores, but there was NEVER a price increase of 50$ in just one generation across all models. So I remember everyone were complaining about Intel's pricing, but now when AMD increases the price by 50$ across ALL models, they're ok with that?? This sounds like fanboyism to me. I mean I do recognize zen 3 is an improvement over zen 2 just like zen 2 over zen+, but not to the point that the 50$ price increase is justifiable.

Not being a fan of price increasing for sure.
But I guess AMD is also tiered of having the budget image.
What they show is solid confidence in the worth of their product. Do I like a price increasement? no - is it justifiable and worth? yes.

Besides it is important for AMD to get some financial buffer to bolster their R&D efforts without always going all in like in a poker game.

I guess next round either Intel is better for the price or they try to undercut it with better Price/Performance. Then AMD has to adjust the price or bring something better....thats just competition. At the end we as consumers will win.

[leipero]

birdie It's absolutely valid to complain about price, however, you need to be objective about it and complain on all fronts - no brand excluded. The issue here is that you started your comment with factually incorrect things:
"Everything as expected however I am not a fan that AMD has OC'ed their CPUs to the absolute limit this time around in order to beat Intel at 1080p by a few percent and by doing so worsened their thermals quite a lot. It would be nice to see all these CPUs with TDP being lowered by 5-20% - that could make them a lot more power efficient and colder."

1. 5000 series have more overclocking headroom than 3000.
2. By all tests I saw, TDP is actually lower in comparison to the 3000, per chip, it's even wider gap when you include performance per watt.

The argument from "Intel fanboys" was always frames per seconds in games at lower details/resolution, because it shows higher latency on AMD CPUs. In games, it's semi-relevant really, if game is poorly programmed and optimized, it will show significant advantage on lower latency and higher single thread performance systems. Intel did ask for arm and leg for their generally inferior products on that basis alone durring Buldozer era even and latter with Piledriver era. Reviewers didn't help the situation at all, by presenting average numbers in benchmarks from games (FPS), often without including wide range of area, hence you got to the situation where reviewers actually recommended dual core Intel CPUs over 8 core Buldozer/Piledriver CPUs on that ground alone, even tho. you had vastly superior product for few extra bucks.

Now, that isn't the case, you trully have a superior product, and if you want lower prices, wait for non-X parts, the thing is, not everyone overclocks their system (being GPU, CPU or RAM), for such people, if they have a budget, that extra cost is worth it. You are probably not that person, but you need to understand that such people exist, and that there's usually a reason why they exist.

[piotrj3]

Besides total RAM (for very edge cases), none of this metrics matter if performance doesn't follow with it. The 3090 has all that more for 15% more performance and 114% higher price.

Quoting Nvidia, 3090 IS FOR EDGE CASES. They literally called it ultimate content creator card. Even in Nvidia presentation 3080 is flagship model not 3090.

[M@GOid]

Besides it is important for AMD to get some financial buffer to bolster their R&D efforts without always going all in like in a poker game.
.

Exactly my thoughts in this mater. Up to now, AMD's success is based in great part on Intel's failure to achieve their roadmap. But everybody knows that will not last forever. They can bounce back next year, or the year after that. They sure have money to burn in R&D. I dare say even more than AMD's entire margin of profit nowadays.

Now is time AMD have to rebuild their R&D and software staff, that they had to cut loose in the near bankrupt years. Look how thin they were, not even bothering to build a temperature driver for their CPU's on the Linux kernel, just last year.

The mater is not if, but when Intel will bounce back, AMD will have the financial means to counter it in short terms. They could not have another Ryzen or Intel's lethargic incompetence to save them again.

[Marco-GG]

9900K, 10700 and 10700K are all better value than AMD CPUs. All top tier CPUs. Very nice attention bias you've got. Congratulations on mastering the AMD marketing school.

10700 and 10700K have as competition the 5800X and the (yet to be announced) 5700X, but you want to compare performance per dollar with the next tier in performance from AMD (for some reason). In any case, taking a look at the TPU benchs graphs for the 5800X it shows almost the same value per USD as the 10700 and 10700k. If your point is: AMD CPUs doesn't give a better bang per dolar than Intel, then you should probably take a look on the 5600X performance per dollar.

Either way, why are you so worried on how we appreciate the value of AMD CPUs? And more over, why everyone that disagree with you must be an AMD fan? You definitely have a mental problem as you're always looking for a fight (or likes). In fact, your first comment in this forum is the same as in TPU forums... I'm almost sure you didn't even read Michael bench and just proceeded to c&p your comment awaiting for the moment to start pointing your finger to anyone that says something not aligned with your opinion. You really are a very immature person.

[muncrief]

Cyrix and Via...those are names I have not heard in awhile. I toyed with a Cyrix 6x86 one time, but did not buy it. Having four kids, you picked carefully - the funds were not there for experimentation. ...
...
I even remember changing EDO RAM in a Gateway 200MHz Pentium MMX computer in my 30s. S**t, I am old.

Oh my goodness f0rmat, we are from the same generation! And I feel you about lack of funds, I had to purchase a 200Mhz 286 CPU in 1989 for a critical project, that fortunately ended up ensuring my future career as a R&D engineer/embedded systems designer, because I couldn't afford a 386 or the newly introduced 486. I would literally sleep at the foot of my computer for 3 to 5 days at a time trying to lay out a Xilinx 3000 FPGA (revolutionary for the time), because they used "simulated annealing" for routing, and it almost always failed. So every hour counted, and when it failed I'd have to immediately get up, adjust the floor plan, and try it again.

I look back at those days quite fondly though. We started with 2 Mhz 8085s and ended up with multi-core CPUs in the GHz, constantly pissing off purchasing departments as we had to buy new test equipment every year at times. However we were on a mission to make the world a better place via technology (at least most of us were), and though we had little idea what we were doing, we had tremendous spirit. I feel very lucky that I came of age in that fantastic era.

[f0rmat]

Oh my goodness f0rmat, we are from the same generation! And I feel you about lack of funds, I had to purchase a 200Mhz 286 CPU in 1989 for a critical project, that fortunately ended up ensuring my future career as a R&D engineer/embedded systems designer, because I couldn't afford a 386 or the newly introduced 486. I would literally sleep at the foot of my computer for 3 to 5 days at a time trying to lay out a Xilinx 3000 FPGA (revolutionary for the time), because they used "simulated annealing" for routing, and it almost always failed. So every hour counted, and when it failed I'd have to immediately get up, adjust the floor plan, and try it again.

I look back at those days quite fondly though. We started with 2 Mhz 8085s and ended up with multi-core CPUs in the GHz, constantly pissing off purchasing departments as we had to buy new test equipment every year at times. However we were on a mission to make the world a better place via technology (at least most of us were), and though we had little idea what we were doing, we had tremendous spirit. I feel very lucky that I came of age in that fantastic era.

Was the first computer that you programmed on a TRS-80 (commonly called a Trash 80)? I learned BASIC on that.

And I think you mean 200MHz 286 CPU in 1999, not 1989. In 1989, The Sun SPARC II Workstation had like a 25 or 33 MHz processor. Then I was working primarily on that and a VAX/VMS 9000 with FORTRAN IV.