Best CPUs (Archive)

Page 5 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.
I don't want to blame you because of the recommendations you made based on what you have learnt from your gaming benchmarks. But I have a strong feeling that your game suite is pretty flawed. I will give you a few examples:

Battlefield 1 - you test that game on DX12 while there are signs that every processor (AMD & Intel) are getting better FPS in DX11 but Ryzen seem to suffer way more from DX12 compared to Intels CPUs. See here: https://i.imgur.com/OFSXbbu.png

Deus Ex - here Intels line up seem to run slightly better on DX12 but AMDs run way better in DX11 compared to DX12. Take a look at this chart: https://i.imgur.com/mrbyQpD.png

Rise of the Tomb Rider: This benchmark is broken (at least in DX12) and you should be aware of it by now. For some reason it doesn't run well with Ryzen + Nvidia GPU but I can't blame you for not testing it with a AMD GPU since they have no high end card yet. But just take a look how "well" Ryzen is scaling up in 1080p compared to 720p to see that something is not right here: http://i.imgur.com/MGW57YI.png

So if you are open to criticism then I just ask you to check your game suite and test the games on different APIs if possible. Whatever API is running better for each processor should be used then in your conclusions.
 


We are testing Tomb Raider in DX11 for the next round of benchmarks. The BF1 results are interesting, I'll see if I can replicate that. Thanks for the headsup.

For Dues Ex DX12, the AMD procs beat the Intel's across the board in our testing.
 


Hi PaulAlCorn

I get it - this is a gaming CPU list (even if its called best CPU) - and I understand your reasons. But I have 2 specific gripes:

- The i3 seems to be obsolote and too expensive now that the Pentium G with threads is out - and I know - Ryzen 3 is not out yet.
- The 4c/4t i5 are now maxed out for online play (demonstrated lots of places) - perhaps for a gaming CPU list you should mention this or directly recommend a stronger CPU with more cores.

So specifically - the i3 and (locked) i5 seems like strange choices - even in a gaming CPU list.

If you haven't fully tested the Ryzen 1500/1600 CPUS you should perhaps hold of recommending anything else until those reviews are done fully. I know I would be kind a miffed to build a locked i5 build for playing BF1 online, for instance.

Anyway - longtime reader of Tom's and I don't ascribe you particularly sinister motives. In general I enjoy you articles and build series.
 



Ok, I hope I'm sufficiently an old enough hand you won't dismiss my comments as being some sort of paid AMD shill.

FIRST let me preface by stating a few things
-I own a desktop and laptop, both sport intel CPUS
-I am an overclocker/computer enthusiast
-I work in IT, and handle a wide range of technology, am intimately familiar with both AMD and Intel product lines.

The problem with your methodology is the lack of "real world" experience.

I'll explain. If you look at pure benches for a modern kaby i3, you'll see it's a respectable gaming cpu. However, anyone using a Kaby i3 on a "real" computer with common "software" installed on it, not a test bench pcs which has just the OS, drivers and testing software, you get a very different experience. I know, I play on enough modern i3s to know they're BARELY functional cpus in OFFICE environments let alone gaming environments. They're crippled by the lack of real cores critically. A fact that amazingly an end user in the real world can experience and see with ease, but somehow escapes most reviewers tests or experience.

-THIS IS THE DISCONNECT- which i've complained about with almost every single review i've seen here and in other places of both cpus and gpus since probably 2012. There is a real and appreciable difference between the testing results in your benches and the end user's experience when they have a real computer in their house, with steam installed on it, an AV installed on it, ms office installed on it, google chrome installed on it, windows 10 doing it's stupid stuff.. the list goes on and on and on for all the things THG and other reviewers DONT test for or even experience in their test rigs, throwing into question the very results they are purportedly are testing for.

-NOW- I'm not a professional bencher nor am I a game tester, I don't believe I can give you guys good advice on how to more accurately demonstrate a real world pc environment with your gaming benches without probably suggesting something which will invalidate my point. But I think my overall point is valid. There isn't a single pentium cpu I'd ever suggest to anyone in the year 2017 to run their office pc, let alone as a "gaming quality" cpu, which raises the question just what on earth do you test?!!! How are you test benches set up that you can't see the criminally poor performance of a pure dual core cpu in this day and age?

This whole probably came into clear focus with me when I moved to this i5-4690k from a piledriver fx8320. My old 8320 was a great overclocker, hit 5.2ghz, I kept it at 5.0 24/7. Had an ssd as a main hard drive. (I need to point that out btw). When I lost that system I replaced it with this DC i5-4690k, also a ssd (both samsung evos, the piledriver was a 750evo i think and this is an 850evo). Now I admit I'm a bit more sensitive to cpu performance that the normal user, and this i5 has been fine (mostly). But I could tell immediately I lost 4 cores when I "upgraded" to this DC chip. Even lightly overclocked (this was a bum chip, never could get it over 4.3, only really stable at 4.2) I've noticed the lack of logical cores (slightly). It does hang briefly playing games that my old fx8320 + r9-280x smashed without issue, this i5+gtx970 seem to stagger from time to time, to straight up hang. Same games, basically same hard drive, same vsync locked 60fps on the same 60hz 1080p monitor, yet the i5+gtx970 gives me a worse experience in some games then the inferior piledriver did?

hard to figure. But then It's not that hard to figure if you think about it a bit. We're talking about a game that smashes 3 cores. The gpus don't matter because they both were able to hit a full solid 60fps in the games in question at the same resolution. but the experience on the i5 is worse (slightly) because it can't manage all the other crap I have going on, on the other 2 screens while I play these games. those 4 cores aren't enough, even though they're undoubtedly much much faster.

Now I've not tried ryzen yet. I've actually purchased a r7-1700, but it's still in it's box because I'm waiting for the itx motherboards (I got a case for it too, but still waiting on the motherboards). SO I don't know how it behaves. But I do suspect all those extra cores will show up in a real world computer pretty obviously like my old piledriver used to. Is this ryzen a sidegrade from my i5, probably, I mean the ipc is closer to broadwell then haswell for ryzen, but the clock speeds are pretty slow, of course my cpu couldn't overclock anyway. I'm not moving to ryzen because I expect it to be faster then what I have, I'm going to ryzen because it will go faster, cooler, lower power consumption and have many many many times the cores then i'm currently using, and I suspect that will show up in a lot of little ways in my daily pc use, which isn't represented by your benching.

As a sidenote: I couldn't with a straight face ever suggest someone purchase a i3 or pentium cpu in this day and age. Furthermore with the ryzen r5-1500(x) cpus sporing 4c/8t I can't imagine any NON-K cpu should be suggested ever again in it's price range. Frankly, I don't think it's actually possible to justify a i~ anything over an r5 right now. Even the i7-7700k. the math actually looks a little worse when comparing my r7 vs the i7-7700k, while the r7 is nice it's just not the same league price/performance, but the r5-1600? as long as that's out there buying a 7700k is just a vanity purchase.

 
this list is somehow true.....in Tom's view a CPU is a gaming CPU 😀
This list is best gaming CPU not "best" CPU. It's just Tomm's miss the "gaming" word in the tile.
(however some CPU here is not the best for gaming)
 
"Intel introduced its Core i3-7100 at the same price as the Core i3-6100, which previously enjoyed a spot on our list of recommendations. Naturally, the -7100 took its place." Very disappointing that it is enough of a justification to get on the list that you liked the old version.
 
Paul Alcorn says: "We haven't yet tested the 1600 and 1400, though we are in the process. As stated, we want to make data-driven statements. If the processors merit it, they will make the list in the next monthly update."

I'm sorry but as a news source you also can't put out an article with the title "Best CPU's for May" and completely neglect to test the Ryzen5 1600 and Ryzen5 1400.

By DEFINITION THAT IS BEING BIASED

It also leaves Tom's Hardware open for litigation by AMD because clearly this is not a BEST CPU list for May because the Ryzen 1600 and Ryzen 1400 is available and was not tested. Because of Tom's position as a reviewer this is not only a biased review it's an incomplete review. You can't publish an incomplete review as a reviewer you loose all credibility. That is what is happening on this board.

You need to update this review with all of the results for ALL of the CPU's currently available to the gaming market. And you need to do it in the next day or two. OR you need to retract this article because it is biased and it is affecting the credibility of Tom's Hardware as a reviewer.

Someone needs to talk to your editor, this should never have been published.

note: If you're going to update the article, then you should address these other sources of bias:
1) The GP bottleneck pointed out in this video which clearly questions your technical conclusions. https://www.youtube.com/watch?v=WlOs_McAVZ0
2) The monetary requirements for the Intel CPU's to overclock them, as you commented on the Ryzen requirements.
3) And you should probably make a comment about the future of gaming as your readers are interested in using their CPU's for games for more than a few months. And we both know that the gaming, and software industry as a whole is moving to parrallel processing, so this should be a large concern for CPU's running future games.
 
Maybe this is finally the death of toms hardware.. the biased reviews have finally caught up with them. Reddit is tearing them apart right now with plans to boycott and what not. Probably be my last visit as well. Good bye Toms Hardware.. was fun but yeah we are done.
 
PAULALCORN -

Ok I will bite on your civil discussion request. I will also give you the benefit of the doubt that you are human and not trying to willfully skew results because I don't (as of yet) intentionally think that you are. I have lurked on your website for ~10 years of my life. You gave my dad and I the recommendations for one of my first builds (which was AMD cpu and Nvidia GPU) when I was just starting college. I use your reviews constantly. I have your webpage saved as one of my favorites at work and I check your site every day for tech news.

I am making my first post today not because I am angry but because I am trying to help you as much as in the past you have helped me. I agree with the other posters here that you are missing the mark "best cpu" as a title but I think I have a better reason why and a fix.

Most "battlestation" pics nowadays are multiple monitor setups (including mine). I don't play a game (I moved out to colorado with my wife) without having a chat setup with my family back in Michigan. My wife and I type our papers, do our taxes and I record and mix on our "gaming" rig. The point is, a single task benchmark is missing where your legacy readers are now at. There is more and more multitasking that is part of a gamer's everyday use because gamers now have families and jobs and we are streaming when we game and watching youtube on another monitor while we game.

I think you can easily fix this buy splitting your article into 3 categories - best gaming cpu line-up, best productivity cpu line-up and best overall line-up. It will be more work but it will accurately help me and my family choose the cpu that will best fit our needs. It will also solve this problem of you missing the fact that my local Microcenter just sold out of their 1600s. There is a reason why. It is the current best CPU around ~ $200 dollars gaming included. I hope this was helpful and I will continue reading your webpage in the meantime but if you haven't tried to include families looking for cpu advice in your webpage then it will be your lost opportunity to grow with those of us who grew up with you.
 
After reading the comments (and the response from Toms calling out questions about methodology) I have some (hopefully) constructive criticism.

For gaming, most gamers game with a web browser open and some sort of voice chat program (i.e. Skype, discord, etc.). I have for a while now thought that these things need to be brought into consideration when doing benchmarks. They tie up CPU resources, often rendering anything with the 4 cores or less crippled when trying to run demanding games.

Second, the article says that it doesn't take into account overclocking. If that were the case then why would you ever recommend the i5 7600k? It doesn't come with a stock cooler (adding to price), requires a Z series motherboard which will cost more, and offers little performance gains over locked i5's with a hefty price increase. Either include overclocking as a variable or don't. Personally I think you should since most people reading articles like this are inclined to overclock. If not, maybe make a second list for overclockers.

Third, I really think using the "we haven't yet tested the 1600" is quite the cop-out. There are many other reputable reviewers that have and praise it for its value. It definitely deserves a spot on the list. If that makes you uncomfortable then put in a disclaimer saying that the recommendation is reliant upon other sources and you will be performing your own tests soon. Also, what the hell is taking you so long to review it!?!

I also think you are avoiding the simple fact that DX12 and Vulvan both are capable of leveraging many more threads than previous API's. You even said in one of your responses that your readers are using CPU's that are 4-5 years old. Can you even pretend that a quad core i5 is going to be enough threads in future games? It barely squeaks by in todays games.

I can honestly see why the Pentium and the i7 are in the list, but to completely leave out the best of the best r5 CPU's seams to be unsettling. I don't believe that you guys are being bought off by intel, but you can't tell me it doesn't look suspicious. I used to look to this site for so much information, but we as readers/viewers have to cross-reference, and other reviewers are seriously praising the r5 1600 and for you to simply ignore it is simply unsettling.

My biggest point though, is the first. Please consider adding in typical background programs that will reflect a proper load that a typical gaming is going to experience.
 
My biggest problem is recommmending the i3 7100. It' probably the worst CPU for price/performance, second only to the i3 7300. The pentium G4560 is easily a much better option as the higher clock speed is never ever worth it on the i3 7100. You either make the jump to the i5 7500 or ryzen 5 1500x or stick to the G4560. The money saved can be put towards something more important, like an SSD or better GPU. It would have made sense to just say:"At the moment of writing, this price point (125-150 USD) offers very bad price/performance and we recommend avoiding it. "

Why recommend something that has such bad value. Benchmark to showcase what I was saying above: http://www.techspot.com/amp/review/1325-intel-pentium-g4560/page5.html (benchmarks on earlier pages).
 
Mr. Alcorn,

All discussion about Intel vs. AMD aside, I still found some problems with your article.

You state that you're "not sure how many more times we can say the list is for gaming". If you don't use a misleading title, you would only have to state it once. Besides taking care of you redundancy problem, it furthermore helps to attract a more relevant audience and prevents the criticism that is now rightfully aimed at you. "Best CPUs" announcing a very different article than "best CPUs for gaming".

You state that you didn't assign the title, yet your name is written under it. It's not apparent who else would be responsible for this article, including the title, apart from you. Do you refuse to take responsibility for the erroneous title? It's okay not to judge a nice bit of online fiction by the proverbial cover, that could even make it more interesting, but there are different rules for a review article. You ought to know if you're willing to put your name on one.

The R5 1600 and 1400 are not included. By that logic, would I be justified in saying that as of May 2017, the R7 1700 outperforms the 7700K in gaming at that price point, simply because I failed to include the 7700K in my roundup? I don't see the article referencing tests run by other people. If there's so many disclaimers about how the scope of this review is limited to gaming only, why isn't there a single disclaimer that the scope of "best CPU" is limited to your selection of tested CPUs? That seems like an oversight that would affect people who aren't fore-armed with knowledge about the current market - the kind of audience this article seems to include.

Hope this is useful, good luck in future.
 
Since it seems you are open to criticism I like to ask you to check your game suite your are using for your benchmarks. There are signs that especially Ryzen suffers in DX12 compared to Intels processors. Let me give you some examples:

Battlefield 1: Overall better results here in DX11 for all CPUs but Ryzen gets way worse results in DX12. See here: https://i.imgur.com/OFSXbbu.png

Deus Ex: Ryzen is getting better results in DX11, Intel is getter slightly higher FPS in DX12. See here: https://i.imgur.com/mrbyQpD.png

Rise of the Tomb Rider: You should be aware by now that Ryzen + Nvidia GPU doesn't run well in this game. Take a look how "well" is scales when comparing 1080p to 720p results here: http://i.imgur.com/MGW57YI.png

So I like to ask you to double-check the games in your game suite and test them on different APIs if possible. Whatever results in higher FPS should be used for your conclusions. This would also be an interesting topic for an article by the way and maybe even would change some of the recommendations you made here.
 
That locked 7500 makes absolutely no sense at $200. Nor does the i3 7100, really a dual core for gaming is a poor choice at any price. The pentium is fine, it can serve as absolute budget chip, but thats all there is in this price range to consider.

7500 already bottlenecks some new games rather badly. It will be fine for most games, but it will bottleneck some newish games, and its only going to get worse in the future. If it was $100 it might make sense, but at $200, its just absurd to reccomend this chip.

The 7600k isnt as bad, at least you can overclock it. But its still overpriced for what it is.

You can get a ryzen 1600 for $220, and have a chip that can process 3 times as many threads as either of those i5s. In any multithreaded workload it will be an absolute slaughter for intel i5s. In games that dont bottleneck on the 7500, it will be a push between the 2. Same games on the 7600k will be a slight nod to intel. In games that bottleneck on the i5s tho it will be smoother on the ryzen.

You can certainly game on the i5s, but you have no where to go; games are already making 4c/4t chips look bad.
 
In a strange turn of events, I now recommend Intel for low end and AMD for high end builds. Actually the only Intel processor I now recommend is the Kaby Lake Pentium. There is really not much reason at all to get i3s, i5s, or i7s from a value standpoint when equivalently priced Ryzens have much more to offer.

Looks like THW Editorial recycled last months's list...shame, shame.
 
Problem: You call this article "Best CPUs of 2017." Then wedged between a big graphic and a bunch of product pictures you say "best gaming CPUs." Too bad it's probably the most important line in the review.

Ryzen's great. Trades blows with or beats Intel in every aspect....except gaming. Heck, even AMD's own graphic* shows that. Not putting "for Gaming" in big bold letters at the top combined with the fact that you don't have a true "Best CPUs of 2017" article (even though this isn't really a gaming-exclusive site) makes you look incredibly biased. Despite the fact that you're technically right.

Right now, I really do think you either need an overall "best bang for your buck" CPU list or a list that's broken down by use case (ie best at xxx price for gaming, for rendering, for other things).

*The one posted at https://twitter.com/AMD/status/860520733029564417
 


Wait didn't Intel just release a statement to not OC Kabylake...
 
It's nice to see so many new members joining yesterday and so many who joined up to five years earlier coming out of the woodwork just to post an attack on Tom's.

That kind of tactic show there is no erudite argument against the substance of the original post or the tests that led to it.

Most of our new [strike]trolls[/strike] members will run off back to Reddit, the world will not fall apart and a vast majority of people will make their personal choice of which CPU to buy on whichever basis they prefer.
 

1) Honestly, it's like saying this article is correct 100% and every comments opposing it is just troll. I made two argument: the title"Best CPUs (of 2017)" is misleading as hell, and this artcile only takes in two account: average FPS and price for gaming, and ignoring multiple smaller, yet when combine together can make a huge difference, aspects.

Got ignored on both LOL
2) People make choice based on influence. No one was born to have all the knowledge in this world. It's article like this that is suppose to give people knowledge, and you just basically said:...nope, this article will not make any difference...Then what's even the point of this article.

But well, it's nice to see a moderator grouping every person who commented as trollers, probably me included.
 
My use of the words "Most of" gets me off that hook. I had the edge over you in that I had to read through all the rubbish which has been deleted.

None of those posts contained any thought through technical arguments against the article but merely criticised Tom's for being in the pay of one manufacturer to attack or ignore another. It was pointed out by our Community Director that it was not the case. He also said there are several degrees of separation between the editorial staff, the head of which produced the article, and any staff who deal with financial issues. That's what keeps Tom's free to all and wholly independent.

Hopefully the flash mob has now gone home and this thread can become technical evaluation of the issues.

Before anyone asks, I'm writing this on an Atom powered netbook but I have AMD chips on my main PC, dual Xeons on my Linux server and an Intel chip on another laptop so I am totally unqualified to support one side or the other. 😀
 


How incredibly insulting and dismissive of feedback of a very misleading headline , combined with a terribly misleading CPU hierarchy posted to your CPU section the same day. Heck Paul Alcorn acknowledged there was misleading in the presentation on both counts, which is a credit to him. Both of those misleading factors vastly favored one brand alone.

You are right, I had not bothered to comment before now, been reading Tom's for decades. I had never been given cause to comment prior by such a misleading headline, and set of recommendations so badly justified in my view as laid out. I have visited your site multiple times daily since the 90s, and have valued it as a reasoned news source for technology. Dismissing feedback because, oh well you have not posted before so your opinion is irrelevant will be bad for business, and lowers my opinion of your organization.

Doesn't this post violate the terms of Tom's forum for waging personal attacks? Terribly disappointed in Tom's Hardware for its article and then mishandled response to criticism. Your comments here have not helped matters for Tom's at a minimum.
 
All accusations aside is any staff member actually going to take a measured approach to any of the actual criticisms?

Yes there is a lot of OTT Hyperbole in this thread by people who may or may not just be here to hate the article, but there are also some very good points being made, points that are not being addressed by anyone from Toms Hardware, who seem far to distracted running around putting out fires....

For example can we have the data on how much load is on these various CPU Compute Threads? how much GPU performance headroom do they have?

If i had the money now for an i5 or a Ryzen 5 1600 which do i get when what i will be doing is keeping it for 4 years but upgrading each year from a GTX 1070 to a GTX 1170, then a GTX 1270 and maybe still have it when i get a GTX 1370? which one will give me the better longevity?

And why is "the best CPU recommendation" now based exclusively on gaming? why can't we have a "Best CPU" in the more true to life rounded sense, i game and do work stuff with my computer, i doubt many actually do use their computer purely for gaming.

Can we actually have some answers to these genuine questions? please?
 
Status
Not open for further replies.

TRENDING THREADS