Be interesting to see what happens to these results when Windows fixes it's AMD scheduler issues, and games optimize for the AMD layout a bit better. If a game can be attached to one CCX unit without bouncing to the other CCX, that will increase performance. Linus showed BFV (I think it was), and it was using cores seemingly at random across the the CPU. Once they assigned affinity to a couple of cores, performance shot up quite a bit and had smoother game play.
4GHz CPU Battle: Ryzen 3900X vs. 3700X vs. Core i9-9900K
- Thread starter Steve
- Start date
Ludak021
Posts: 775 +590
The problem here is that out of 5% of people that are in the market for any of those two CPUs, 90% don't care about any of those results, they just play games and otherwise use PC in a manner that 10 year old CPU can satisfy. I could give you statistics about how over 80% of Americans use mobile phones and tablets for everything (which means that they don't even know what things in your list are), and how the rest of the world doesn't really care unless they are in the industry that uses Handbrake (in witch case they are broke) or other applications from your list. In that case, they just get the threadripper or xeon, not these toys.
Evernessince
Posts: 5,469 +6,158
Actually this is not true at all. More people still use their PCs for office applications and professional work then gaming. Gaming is the 2nd biggest use case for PCs. The productivity tests like compression and decompression apply to nearly everyone, gamer or otherwise. You can bet that if a CPU is doing well in nearly all applications that they are going to see a more responsive system in general.
FYI not everyone jumps straight to threadripper or xeon. Not to mention the additional IPC these have over those server class CPUs. There are many workflows which benefit from both cores and IPC but they don't need or want a ton of cores with less IPC. Video editing and production for example.
amstech
Posts: 2,661 +1,824
@Evernessince, even though we dispute I got nothing but love for you. I respect your opinion, angle, input and thoughts on everything hardware and always will. It's no secret your highly intelligent, who just likes AMD the same way I like Intel.
I admire that, I truly do.
I also admire anyone else in these last few articles about Ryzen's latest stuff and thanks to Techspot for getting great results that handily cover just about everything. I apologize to folks about my spirited take on the 8700K/9700K against the 3700X.
My take on the 8700K/9700K vs the 3700X in games is not an attempt to discriminate or demean its wonderful overall performance for the price, and yeah, it games pretty damn well too, I never said it didn't. But there is a difference in that specific range for a specific purpose, and its a popular one. Never meant anything else by it.
As far as the ratio of folks who own a desktop PC at home and what they use it for? Thats a good question.
The PC gaming industry is powered by the millions and millions of dollars spent on gaming cases, RAM, mice, monitors, keyboards, speakers, hardware and heck, even gaming chairs. I won't argue about what percentage of folks use their PC to game, but you can bet a significant amount of folks who build or buy a performance desktop do it to game.
I admire that, I truly do.
I also admire anyone else in these last few articles about Ryzen's latest stuff and thanks to Techspot for getting great results that handily cover just about everything. I apologize to folks about my spirited take on the 8700K/9700K against the 3700X.
My take on the 8700K/9700K vs the 3700X in games is not an attempt to discriminate or demean its wonderful overall performance for the price, and yeah, it games pretty damn well too, I never said it didn't. But there is a difference in that specific range for a specific purpose, and its a popular one. Never meant anything else by it.
As far as the ratio of folks who own a desktop PC at home and what they use it for? Thats a good question.
The PC gaming industry is powered by the millions and millions of dollars spent on gaming cases, RAM, mice, monitors, keyboards, speakers, hardware and heck, even gaming chairs. I won't argue about what percentage of folks use their PC to game, but you can bet a significant amount of folks who build or buy a performance desktop do it to game.
Evernessince
Posts: 5,469 +6,158
I really do not deserve the kind words. I appreciate the gesture.
So long as Intel and AMD are competitive, that's good enough for me. I want to see both push the envelope of what's possible as we need them both to do good.
Ludak021
Posts: 775 +590
Look, FYI (I hate that term, it means nothing and accomplishes nothing), not everyone with Xeon farms will jump ship over even 10% better unzipping time (which they don't even need). Get some sense about what businesses that make thousands per day using computers are going to use, when they are going to upgrade and why and you might give one big "FYI" to yourself. These CPUs in the article are TOYS. Why would anyone care for 7-zip unzip time? Is it 60 seconds? 120 seconds? 10 seconds? It's irrelevant, it's sparsely used and the time is low on any of those I just listed.
"Doing nearly well in all applications" so does the other one. Would you beat yourself up for having a 7700K and not 9900K or 3900x if your "any application" comes up twice a week? Or even once per day? Be real.
Evernessince
Posts: 5,469 +6,158
Why would anyone care for decompression time eh? You don't seem to realize that when you deal with a certain type of file all day long and you now have a product that will reduce the time from 120 seconds to 90 seconds 100 times a day that's a huge time savings. Or do you not realize how prevalent compression / decompression is in our everyday world? Images, videos, Audio, game textures, most rich media formats are compressed.
Give that the 3900X is a whopping 342% faster in decompression then the 7700K, I think you are vastly misstating the performance of some of the products here. In fact it absolutely trashes the 7700K in application benchmarks which you can see here.
https://www.techspot.com/review/1869-amd-ryzen-3900x-ryzen-3700x/
It's not just "a application", it beats the 7700K in everything including a majority of games. A good chunk of those applications are in the 3900X's favor by a huge margin as well, as demonstrated above.
So no, the 7700K is nowhere close to the 3900X in applications. Even in games it is starting to fall behind far weaker 8 core offerings from AMD.
Ludak021
Posts: 775 +590
Yea, I think we have to disagree. No one decompresses anything 24/7 save for you in your head. Sorry. Having things in zip inherently slows performance on anything. After this, you just completely miss the point because you didn't get the first one. Oh well...Maybe you just need a new ssd or HDD to save on time with decompressing files? I think investing 50$ would increase performance for you by 1000% don't you think?
Evernessince
Posts: 5,469 +6,158
Performance is better or vastly better in EVERY application. No one ever said anything about compressing / decompressing 24/7. It's called an example, which was used to illustrate the difference. Compressing 100 archives a day of 90 seconds each isn't what I'd call "24/7" as you put it either. In fact it's exactly 150 minutes. It's a good example because my previous workplace required us to compress and encrypt files that were being sent over the internet and that would typically be much greater then 100.
I think you just like making hyperbolic statements that are incorrect. Why would anyone want to get your point when it's based on wildly inaccurate information?
Ludak021
Posts: 775 +590
Why would anyone want to get a take on these benchmark results solely based on everything you like and not me? I like games. AMD is not as good as intel. What now? Why is your argument better than mine? I am eager to hear it. (you asked for this) Games are "application" and well....I could probably name a 100 of those applications, and you couldn't name 100 of "your kind" of applications (that no one uses in this segment anyway). And also, you refuse to respond to other point's I brought, because they don't suit you. Such as, 5 year old xeon is 2x cheaper than new amd and as fast. Why would I upgrade? I don't play games obviously, I just unzip files, do handbrake and stuff, there is no improvement for me, I already have 32 threads xeon.
Last edited:
The 3600 wasn't mine and the owner was still tinkering with it. My point was that the 3600 is good if you plan on combining a cheap mobo and the stock cooler, it's basically a stock 8700k for half the money. At least in this game
Tristann
Posts: 6 +30
You came to this forum and you sound like you were looking for an excuse for your Xeon. Since you have Xeona you do not play much and you work in applications and you can not deny the fact that AMD far ahead of Intel in this field, and in terms of one thread and many threads. Work on such a 5-year-old Xon with a small clock and weak IPC must be a punishment, most operations when composing static images or video must run sluggishly and inefficiently compared to Ryzen 3000, the difference can be as high as 100%, Only when Rendering such a 16 core Xeon from before 5 years can approach the efficiency of 8 core Ryzen 3000. "AMD is not as good as intel." - so this sentence should sound, Intel is currently not as good as AMD.
urbanman2004
Posts: 172 +56
3700X FTW ;-)
So...all AMD/Intel fanboys - stop thrusting in benchmarks. I tried 3 times to change platform from Intel/NVidia to AMD/Radeon. My last was 2700X + Vega 56. Three times I lost a lot of money to get worse performance. I'll tell You why. I'm an VBA programmer so performance in running Excel macros is the most important to me. So if Ryzen is fastest than Intel - it should do better in Excel too...it is simple. Fastest chip should have more raw power in not-special prepared benchmarks.
What the real facts are? I have a macro that analyze big data file and get some data from it and write to new Excel spreedsheet with (finally) 3404 worksheet in it. These are the times in secons for few CPU:
Intel X5667 + DDR3 1333 triple channel: 6h35m (very old X58 generation)
Intel 6600K + ddr3 1600 dual channel: 3h30m
Intel 7700K + ddr3 1600 dual channel: 3h05m
AMD 2700X + ddr4 2666 dual channel: 3h25m
AMD 2200G + ddr4 2666 dual channel: 4h11m
And remember that Intel still will be faster in scientific workloads becouse of wider AVX512 support. Even now when AMD improves Ryzen 3 AVX256 execution in one cycle (not 2 cycle = 2 x AVX128) Intel is faster in AVX256. And do it on relatively slow and cheap DDR4 2666 modules.
So 7700K is now 3 generation before new Intel Core X and still beat AMD Ryzen 3rd generation. I don't even think how fast cheap I3-9350K will be in Excel, games and so on, even on 4 high frequency cores.
Conclusions - First, low latency monolithic chip will be always faster than multichip modules in gaming and programming. Use Ryzen for video encoding and renderings where core count is more important and memory latency has no impact for this type of workloads. And thats all. I remember that AMD on it's presentation try to show that Ryzen is fastes than Intel in Excel workloads - it was just specially prepared and I hate AMD for that. It's the same situqation when NVidia cuted off anizo quality in drivers to match old Radeon speed. I hated that in that time and hate it now. Have a nice day everyone.
Ah - once more - 2200G is very, very slow and integrated GPU is not a reason of that. On external GPU and disabled internal speed changes by 1-2% margin. WoT is making all 4 cores at 80% utilisation and still are big fps drops in action. Don't get it unless You not only browsing internet.
What the real facts are? I have a macro that analyze big data file and get some data from it and write to new Excel spreedsheet with (finally) 3404 worksheet in it. These are the times in secons for few CPU:
Intel X5667 + DDR3 1333 triple channel: 6h35m (very old X58 generation)
Intel 6600K + ddr3 1600 dual channel: 3h30m
Intel 7700K + ddr3 1600 dual channel: 3h05m
AMD 2700X + ddr4 2666 dual channel: 3h25m
AMD 2200G + ddr4 2666 dual channel: 4h11m
And remember that Intel still will be faster in scientific workloads becouse of wider AVX512 support. Even now when AMD improves Ryzen 3 AVX256 execution in one cycle (not 2 cycle = 2 x AVX128) Intel is faster in AVX256. And do it on relatively slow and cheap DDR4 2666 modules.
So 7700K is now 3 generation before new Intel Core X and still beat AMD Ryzen 3rd generation. I don't even think how fast cheap I3-9350K will be in Excel, games and so on, even on 4 high frequency cores.
Conclusions - First, low latency monolithic chip will be always faster than multichip modules in gaming and programming. Use Ryzen for video encoding and renderings where core count is more important and memory latency has no impact for this type of workloads. And thats all. I remember that AMD on it's presentation try to show that Ryzen is fastes than Intel in Excel workloads - it was just specially prepared and I hate AMD for that. It's the same situqation when NVidia cuted off anizo quality in drivers to match old Radeon speed. I hated that in that time and hate it now. Have a nice day everyone.
Ah - once more - 2200G is very, very slow and integrated GPU is not a reason of that. On external GPU and disabled internal speed changes by 1-2% margin. WoT is making all 4 cores at 80% utilisation and still are big fps drops in action. Don't get it unless You not only browsing internet.
lol ya if you go back and read this again there is alot of fan boys on this site and one track minded users that really only play games and do nothing else.
I don't think it's a con for the vast majority of gamers, which are those with any kind of budget at all on a graphics card and/or don't specifically play lots of fast twitch shooty shooters and/or play at 4k and/or feel like any diminishing returns at ultra high framerates don't outweigh the flexibility of more power for other tasks/platform upgradeability, etc. I think that must be why AMD is outselling Intel so significantly with knowledgeable DIY enthusiasts.
To be fair, the strength of an architecture isn't just the IPC but CLOCK speed is equally important. The design of the architecture needs to take both into account. Node or process is a part of the equation too certainly, but Intel has done a great job with it's 14nm+. That said, Intel groupies place too much importance on the diminishing returns of extra frames at high framerates and speak as if Intel is the only way to go for gaming with no qualification.
EEatGDL
Posts: 770 +501
Yep, they achieved that with the SK775 Pentium 4s and Pentium Ds architecture, then dialed the clocks way back with Core 2 Duo. Reason? Netburst architectures having north of 35 pipeline stages and Conroe going back to 8. That was an architecture limitation with same manufacturing process. So why are the GHz relevant when Netburst was great for marketing and not much else?
In the current AMD situation it doesn't seem to be tied to architecture but manufacturing process (node). Go take a look at Intel's current 10 nm offerings and compare them clockspeed and TDP-wise to realize things aren't any better there (SPOILER ALERT: they are dialing back the clocks, with lower core count, and same or higher TDP; compared to 14 nm offerings). It will be even a lot more evident when you see the server offerings, but don't hold your breath: that will take a long time before we can resume this conversation.
By the way. Techspot, as I am sitting here in comments..it seems as if something from your site is spinning off extra Chrome threads for something. It's happening on this and another computer I have. Was getting some quick odd little messages in the bottom left corner of the screen. "waiting for spotx, connnatix, etc, etc, hard to make them out"
I wouldn't make that assumption. Also, Intel might even decide to nix the ringbus and take a slight hit to gaming, if it means enough advantages in other areas.
No it doesn't. Many other factors to consider.
Last edited:
The majority don't play shooty shooter games at low res on a 2080ti, either.
Last edited:
Most DIY enthusiast gamers are buying Ryzens, so they must see some additional value in Ryzens outside of the gaming crown at low resolutions with very high end graphics cards.
Last edited:
Similar threads
- Locked
Latest posts
-
AMD Radeon Vega 64 Revisited: Did It Age Better than GTX 10 Series?- Mastodonrules replied