Any point to Intel X-Series CPUs?

[CAT-THE-FIFTH]

Another use case for Intel is old RTS or old Bethesda games,etc which won't be updated for newer CPUs,and use only a few threads,and like the ringbus Intel CPUs. But then you could argue a 9600K would be fine for that,and I expect Ryzen 4000 might not even make those niche case scenarios a win for Intel.

 

[NoPost]

My last 3 rigs were/are :
X299 with 7960x > Z390 with 9900K and now TRX40 with 3970x which completely destroys Intel's X299 offerings.

For me, anyone who even considers buying into the current Intel HEDT platform must by mildly retarded.

 

[Gavin786]

Ok, so you guys had me pretty much convinced to go with AMD and forget about buying Intel.

Then I check out this : https://cpu.userbenchmark.com/Compare/Intel-Core-i9-9900KS-vs-AMD-Ryzen-9-3950X/m929964vs4057

Basically its saying that for any use case except in which I have 8 cores fully maxed out in usage the Intel is going to be the better CPU? For all performance scenarios upto and including 8-core performance, the Intel is the better chip?

Am I interpreting this right ?

Is anyone actually doing something that needs 9 cores/18 threads? I usually run a few VMs and my laptop 4core/8thread CPU seems to be fine(with my use cases anyway), I did buy a hefty chunk of RAM, but CPU seems like not a bottleneck.

Mostly I am going to be gaming/developing. And potentially dual or triple-boot system, depending on whether I can get away without windows. Which is why compatibility is a factor for me. I have not looked into whether I can get away with an AMD CPU for a hackintosh, but I know the Intel 9900KS is compatible.

I do like the PciE Gen 4 on the SSD, but this seems to be the only solid advantage the AMD is going to have for normal usage scenarios. Are there any games or software that seriously take advantage of this massive parallelism? Mostly games run on 2 cores at most, as far as the ones I play goes, which only 3AAA game is Elite pretty much. KSP is something else I play and is single-threaded performance is most important for it?

What am I missing, am I seeing the numbers wrong?

I am not trying to argue btw, just dont know its why Im asking, I dont have a position, I am trying to find out.

And thank you for everyone who has taken the time to reply so far.

 

[LePhuronn]

You are interpreting the charts correctly. The problem is the charts are ******* trash, almost to the point of outright lies.

Userbenchmark has been heavily messed with to pander to Intel. Never, ever, ever use it.

 

[Gavin786]

You are interpreting the charts correctly. The problem is the charts are ******* trash, almost to the point of outright lies.

Userbenchmark has been heavily messed with to pander to Intel. Never, ever, ever use it.

I didnt know that.

What benchmarks are the correct ones, are they available?

How, specifically, are userbenchmark fiddling their results? I know in the past(coming up 20 years ago now lol), there was Intel Performance Primitives which was a set of libraries that did a lot of mathematical functions, and was great at speeding up certain core inner-loop critical stuff. It turns out that the library was deliberately designed not to work with AMD CPUs, reverting to a much slower version. And when it was cracked/modified to run on AMD it performed as good as equivalent Intel. TLDR; it was synthetically tampered with to make AMD run worse. So if this type of thing is used in a 'benchmark' then I can see that skewing things a lot, and I know it is a trick Intel pulled in the past.

I see the CPU clock cycle numbers there, and I am having a lot of trouble understanding them. It says AMD runs at 4.7 GHz, somewhere it says a lot lower than that, and the overclockers said they could only get 4.3(?) which is lower than the advertised speed? I dont know what is going on. For the Intel, I got one here, and I overclock it to 42x on the multiplier and it will run quite happily at that speed ~4200 on all the cores according to cpu-z(remember I dont know jack **** about serious over clocking, I am just using the tools with the msi and not even the bios). What is going on with AMD clock cycle numbers? What is going to be the actual real clock-speed I can expect?

It seems tho that the 9900KS can reach lets say 5.1 overclocked. People, even amateurs can get this consistently. I certainly would expect this if I buy. And this number seems easy to understand, no fiddling with it.

Let us assume for a moment that these figures are correct and the Intel is running at 5.1 and the AMD at 4.7, all CPU across all cores. This would mean that for the AMD to match the Intel, it would need to do more per cycle than the Intel, or else have something else going on, on the board say, that would compensate for lower clocks. Performance up to 8-core, one would assume, would be better on the Intel, especially in applications which are not parallelised well.

Anyway, I guess if the userbenchmark score is wrong, are the right scores anywhere? How do clock cycle numbers compare, are the CPUs like-for-like on clock cycle, or does one have more effective speed-per-cycle? I know clock cycle is not the be-all as some instructions take more than 1 clock, some can be parallelised, etc.

Thanks guys, I know I am probably annoying some of you by now with my n00by questions.

 

[Tetras]

It's really best to compare across a range of different workloads that suit your specific needs, userbenchmark is reasonable for comparing different generations, but it's not so great for use as a comparison tool within the same generation and I'd never use it in isolation.

It doesn't just apply to them though, for example: the general assumption for a long time was that Intel CPUs are superior for games, but now that games have greater multi-threading support, older 1st and 2nd gen Ryzen are cruising way ahead of the same generation Intel i3 and i5 CPUs. Some reviews are still done with much older (mainly DX 11) games and that gives a distorted and out-of-date perspective of gaming performance.

Generally speaking, with Intel you now pay a lot more than AMD for less cores/threads and for slightly higher lightly-threaded performance. With 3rd gen Ryzen their performance is actually superior at the same clock speed which is pushing Intel to run the architecture way beyond optimal efficiency. I'm more familiar with this at the low-end, but apparently it still applies at the high-end too:

"Is the 9900KS Special addition an exciting processor or not? That is a question that I have been asking myself. The answer to that is more complicated then I would have thought. On a personal note, I like it, purely from a gaming point of view squeezing out that last bit of perf out of your graphics card. And even there there are restrictions, let me quickly rephrase; this processor will squeeze out every last frame and is the faster processor for enthusiast-class gaming. Why do I highlight enthusiast-class you wonder? Well, because you will only see these differences with graphics cards running 600 USD and upwards, you need to be free from GPU bottlenecks, and realistically 90% of you guys are GPU bound, and there your processor matters far less as the game runs as fast as your graphics card can cope with. But comparing apples to oranges, yes this is the fastest gaming processor your money can get you. That money is the second thing of interest, the 9900KS is just over 500 bucks. For the same amount of money, you can get the 3900X with 12c/24t. It will rip a hole on the 9900KS in pretty much anything single-threaded and multithreaded thanks to the four more cores. But enthusiast gaming-wise (yes, here we go again) the 9900KS will offer the win here if you have a fast enough graphics card (I can't stress that enough). Pricing wise I feel it's a fair amount of money though, at roughly 65 bucks per core, considering what it offers: leading performance for enthusiast performance gaming.".

https://www.guru3d.com/articles_pages/intel_core_i9_9900ks_processor_review,26.html

It has already proved the case that a decision to go Intel (less cores/threads for slightly higher lightly-threaded performance) fared badly against 1st/2nd gen Ryzen, so I'd be weary about making the same decision when multi-core support seems to be taking off now. (There's also the utility of those extra cores, even if only 8 are being used in a game)

 

[humbug]

I know AMD are currently slightly better multi-core performance than Intel at the moment.

For me reliability is an important concern, and my experience with AMD in the past is that there has always been the odd compatibility issues, and in the case of their graphics card, bad driver support, games not working, etc, etc.

And that is something that as time gets on, frankly, I just cant be bothered with.

I am happy to take a hit in price or performance to get something that is a bit more rock-solid without compatibility issues.

That is why I am looking at Intel and am loathe to consider AMD, even though they might be slightly better right now.

So I would kind of still like an answer to my questions regarding the X-series CPUs, what is the point of them, should I really consider them over just the KS ?

I would like to maybe try a hackintosh(lol and I was complaining about compatibility issues), so I think I need one K or KS with integrated GPU. And anyone who knows about this topic, advice would also be greatly appreciated.

I probably will wait later in year till buying, as I dont want to buy before the next gen Nvidia graphics cards come out, but am itchy to buy as soon as I can.

I know AMD are currently slightly better multi-core performance than Intel at the moment.

Slightly? mate its a blood bath!!!! AMD's Best is up to three times as fast as Intel's best.

https://www.anandtech.com/show/15483/amd-threadripper-3990x-review/4

 

[Orange Nexus]

Ok, so you guys had me pretty much convinced to go with AMD and forget about buying Intel.

Then I check out this : https://cpu.userbenchmark.com/Compare/Intel-Core-i9-9900KS-vs-AMD-Ryzen-9-3950X/m929964vs4057

Basically its saying that for any use case except in which I have 8 cores fully maxed out in usage the Intel is going to be the better CPU? For all performance scenarios upto and including 8-core performance, the Intel is the better chip?

Am I interpreting this right ?

Is anyone actually doing something that needs 9 cores/18 threads? I usually run a few VMs and my laptop 4core/8thread CPU seems to be fine(with my use cases anyway), I did buy a hefty chunk of RAM, but CPU seems like not a bottleneck.

Mostly I am going to be gaming/developing. And potentially dual or triple-boot system, depending on whether I can get away without windows. Which is why compatibility is a factor for me. I have not looked into whether I can get away with an AMD CPU for a hackintosh, but I know the Intel 9900KS is compatible.

I've been wanting to try hackintosh, too, and i found this guy in yt doing such on an X370 platform last year. This is his latest i think on a X570.

https://youtu.be/0wACjS8XRLo

 

[jigger]

Ok, so you guys had me pretty much convinced to go with AMD and forget about buying Intel.

Then I check out this : https://cpu.userbenchmark.com/Compare/Intel-Core-i9-9900KS-vs-AMD-Ryzen-9-3950X/m929964vs4057

Basically its saying that for any use case except in which I have 8 cores fully maxed out in usage the Intel is going to be the better CPU? For all performance scenarios upto and including 8-core performance, the Intel is the better chip?

Am I interpreting this right ?

Is anyone actually doing something that needs 9 cores/18 threads? I usually run a few VMs and my laptop 4core/8thread CPU seems to be fine(with my use cases anyway), I did buy a hefty chunk of RAM, but CPU seems like not a bottleneck.

Mostly I am going to be gaming/developing. And potentially dual or triple-boot system, depending on whether I can get away without windows. Which is why compatibility is a factor for me. I have not looked into whether I can get away with an AMD CPU for a hackintosh, but I know the Intel 9900KS is compatible.

I do like the PciE Gen 4 on the SSD, but this seems to be the only solid advantage the AMD is going to have for normal usage scenarios. Are there any games or software that seriously take advantage of this massive parallelism? Mostly games run on 2 cores at most, as far as the ones I play goes, which only 3AAA game is Elite pretty much. KSP is something else I play and is single-threaded performance is most important for it?

What am I missing, am I seeing the numbers wrong?

I am not trying to argue btw, just dont know its why Im asking, I dont have a position, I am trying to find out.

And thank you for everyone who has taken the time to reply so far.

Threadripper VM machine maybe? Run all the systems at the same time?

 

[LePhuronn]

What benchmarks are the correct ones, are they available?

Actual product reviews by the tech press are more reliable, but even then you can see some bias with certain writers. Personally I'd just search for Ryzen 3000 reviews and scan over the results for about 5-10 different sites.

How, specifically, are userbenchmark fiddling their results?

In a nutshell, any metric which has given performance parity or superiority to Ryzen has been downgraded in importance to the overall CPU's ranking. The biggest shift is the value of multi-core performance was dropped a lot, and the value of single-core performance was elevated a lot, which gives you utterly ******** situations where a Pentium is suddenly a "better" CPU than the 3900X because it can run 7zip faster.

https://www.tweaktown.com/news/66768/userbenchmark-adjusting-cpu-rankings-pandering-intel/index.html
https://forums.anandtech.com/thread...-the-score-weights-in-favor-of-intel.2567962/
https://www.reddit.com/r/buildapc/comments/chco8h/userbenchmark_should_no_longer_be_used_after_they/


This does also screw over Intel's bigger core count CPUs too, suddenly rendering an 18 core Skylake X inferior to a Pentium, but that doesn't matter too much because the average visitor to userbenchmark isn't looking at HEDT: they want to play games, they want to watch YouTube and maybe do some creative stuff and are essentially told that Ryzen is ****.

And this all happened shortly after Ryzen 3000 launched and battered Intel to a bloody pulp.

 

[sideways14a]

This is why Userbenchmarks appalling results are more of a big deal than many are prepared to accept.
I have heard of a lot of folk who know virtually nowt about computers go to this garbage intel shill site and come back quoting ridicules numbers for rubbish intel chips v's AMDs.

Intel know what they are doing and while we dont know if Userbenchmarks have been paid to lie or are simply stupid having them buff up intels products compared to AMDs will secure them a lot of noob sales.

 

[Gavin786]

Threadripper VM machine maybe? Run all the systems at the same time?

Id love to do that, it would be the perfect solution. Problem is I have had right trouble virtualizing Mac OSX in the past. It is extremely slow, the graphics does not virtualize at all well, and the graphics card seems just not to be used full-stop. Running direct on the hardware is night and day. However, I would be wise to have another go at doing this before I commit to buying a more expensive board for a hackintosh. I have never been in the situation where I needed a top-level hypervisor like vsphere to use the graphics card in the VMs. It might be worth looking into if this works nowadays, or the graphics card is still virtualized. Thank you.

 

[~>Dg<~]

is there any point was the question. the simple answer is yes despite the debaiting. for gaming it will be quicker than anything out now on AMD. quite a bit quicker than a AMD 3900 for eg you just lose 2 cores. thing is most people wont use 8 or more cores anyway. so for most people it wont matter.

 

[jigger]

Id love to do that, it would be the perfect solution. Problem is I have had right trouble virtualizing Mac OSX in the past. It is extremely slow, the graphics does not virtualize at all well, and the graphics card seems just not to be used full-stop. Running direct on the hardware is night and day. However, I would be wise to have another go at doing this before I commit to buying a more expensive board for a hackintosh. I have never been in the situation where I needed a top-level hypervisor like vsphere to use the graphics card in the VMs. It might be worth looking into if this works nowadays, or the graphics card is still virtualized. Thank you.

Nvidia are a problem in a virtual environment, but passing through a Radeon card is pretty well documented.

 

[LePhuronn]

I don't see any debate, Dg, rather a unanimous opinion that Intel is inferior in all but the extreme FPS gaming niche. Are you reading posts through blue-tinted goggles again?

 

[RavenXXX2]

Nothing else to say but they were
Dead On Arrival

 

[Gavin786]

Thanks to all the guys who posted links and gave their valuable opinions on the topic. I have read all and gave consideration.

So I have looked into things and it certainly seems that userbenchmark are not playing fair and altered their scoring mechanism when AMD Ryzen came out so that Intel would continue to come out on top of their rankings. Which is very unfair. However, this does not necessarily mean their results are wrong for the typical use-case of Gamers/Devs. As far as I can see, having looked into it now, that AMD only wins out when doing rendering for YouTube videos, or some extreme networking testing where a dozen VMs are in use at the same time. Also AMD wins in synthetic benchmarks in which more than 8(9?) threads are simultaneously utilised or there is no thread limit. Which, given how rare these scenarios are going to be for me, is seriously not worth a drop of 15%-30% on regular day-to-day performance. I dont do any rendering of 3D graphics and if I was, I could do it while I am asleep if it was going to hog my CPU for hours at a time.

I checked the gamernexus video @RavenXX2 kindly provided(thank you), and it does seem that the Intel chips(esp 9900K(S)) beat the AMD in every real-world scenario. Sometimes there is even a 30% improvement in games for the Intel chip vs the AMD I would want to buy (3950x is my AMD shortlist winner).

Seems like, even though they played a dirty trick on AMD, the userbenchmark score is consistent with the results gamersnexus got and reflects most people's real-world use-case and experience.

I know this conclusion is probably going to annoy a lot of people, but the numbers is usually what I go with over opinions, and they are all saying the Intel is better, except in the most extreme and rare circumstances/synthetic benchmarks utilising 9+ threads.

I have been the victim myself of reviewers skewing results in favour of their favourites, and ignoring unique features/advantages of my products, so I know how miserable and unfair this is, and I do keep this in mind. And I do want to slant things a little bit in AMD's favour, esp based on the opinions above, and the misbehaviour of userbenchmark, but the figures just dont support this, and I cant make them fit the scenario where AMD will be a better choice for me. The only way this could change is if games suddenly started supporting massive parallelism, which really does not seem like it will happen in the next 3 years or so.

So conclusion seems to be that if your main thing is video rendering/3d graphics rendering and certain editing tasks, then the AMD 3950x is the way to go. For every other thing, esp gaming, the Intel 9900KS is the best chip.

People seem to be saying that the numbers are wrong and that AMD is the better chip, but I have yet to see any cogent argument that supports that position. I care about PCIe gen4 and the faster SSDs, but I dont think this is enough to sway me.

All this really begs another question though : Are Intel/AMD likely to release another chip this year? Is there any good reason to buy now, or is waiting till later the best move?

 

[sadbuttrue]

Thanks to all the guys who posted links and gave their valuable opinions on the topic. I have read all and gave consideration.

So I have looked into things and it certainly seems that userbenchmark are not playing fair and altered their scoring mechanism when AMD Ryzen came out so that Intel would continue to come out on top of their rankings. Which is very unfair. However, this does not necessarily mean their results are wrong for the typical use-case of Gamers/Devs. As far as I can see, having looked into it now, that AMD only wins out when doing rendering for YouTube videos, or some extreme networking testing where a dozen VMs are in use at the same time. Also AMD wins in synthetic benchmarks in which more than 8(9?) threads are simultaneously utilised or there is no thread limit. Which, given how rare these scenarios are going to be for me, is seriously not worth a drop of 15%-30% on regular day-to-day performance. I dont do any rendering of 3D graphics and if I was, I could do it while I am asleep if it was going to hog my CPU for hours at a time.

I checked the gamernexus video @RavenXX2 kindly provided(thank you), and it does seem that the Intel chips(esp 9900K(S)) beat the AMD in every real-world scenario. Sometimes there is even a 30% improvement in games for the Intel chip vs the AMD I would want to buy (3950x is my AMD shortlist winner).

Seems like, even though they played a dirty trick on AMD, the userbenchmark score is consistent with the results gamersnexus got and reflects most people's real-world use-case and experience.

I know this conclusion is probably going to annoy a lot of people, but the numbers is usually what I go with over opinions, and they are all saying the Intel is better, except in the most extreme and rare circumstances/synthetic benchmarks utilising 9+ threads.

I have been the victim myself of reviewers skewing results in favour of their favourites, and ignoring unique features/advantages of my products, so I know how miserable and unfair this is, and I do keep this in mind. And I do want to slant things a little bit in AMD's favour, esp based on the opinions above, and the misbehaviour of userbenchmark, but the figures just dont support this, and I cant make them fit the scenario where AMD will be a better choice for me. The only way this could change is if games suddenly started supporting massive parallelism, which really does not seem like it will happen in the next 3 years or so.

So conclusion seems to be that if your main thing is video rendering/3d graphics rendering and certain editing tasks, then the AMD 3950x is the way to go. For every other thing, esp gaming, the Intel 9900KS is the best chip.

People seem to be saying that the numbers are wrong and that AMD is the better chip, but I have yet to see any cogent argument that supports that position. I care about PCIe gen4 and the faster SSDs, but I dont think this is enough to sway me.

All this really begs another question though : Are Intel/AMD likely to release another chip this year? Is there any good reason to buy now, or is waiting till later the best move?

Why are you comparing a 3950x to the 9900KS? A more useful comparison would be a 3700X AND a 4700x on release, You can make a keyring from the 3700X, still be in profit and have a faster system than the 9900KS.

 

[Toasty]

Thing is 9900KS is really quite expensive now as rarer by the minute and it’s still only an 8 core and the end of the day.... No way Zen2 is 15-30% behind outside of total cherry picked outlier cases. In games maybe 5% or so. I have a 3700x, 5.3 9900KS, 3950x and a 3970x all water cooled with overkill rads and if we’re talking just gaming I could not tell the difference between any of them as I play either 3440x1400 or 4K on 2080TI.

 

[Troezar]

I suspect this thread is thinly veiled excuse to promote an expensive EOL Intel CPU. I wouldn't spend too much time trying to convince anyone. One attempt is all they get from me then it's caveat emptor.

If you've done your research the answer is obvious. I bought an expensive motherboard for a reason, the 4950X. Try that with LGA1151...