Modern day CPU reviews and what is wrong with them.

[mattyfez]

But Surley if you're building a gaming machine, then game benches are the most important.

If you want a multi tasking power house, then that's a different horse for a different course.

Or a combination of both in which case you weight up the pros and cons with bias to your personal priority?

 

[N19h7m4r3]

But Surley if you're building a gaming machine, then game benches are the most important.

If you want a multi tasking power house, then that's a different horse for a different course.

Or a combination of both in which case you weight up the pros and cons with bias to your personal priority?

That's the way I see it, and how I came to buy my 5820K back when x99 launched.

As it stands, if I was looking to build a similar system today, I'd be daft not to go for a RyZen 8 core system given the value for money for productivity and gaming.

What is rare is also seeing RyZen compared to the x99 setup in gaming though. You only rarely see the 6900K in a graph, but not much else done.

I found this video really interesting. 1600X at 4Ghz vs 6800k at 4.2Ghz in both gaming and productivity.

 

[DragonQ]

In which case there is no or very little difference between these CPU's, all these reviews, Toms Hardware, Tech Power Up ecte.... have the Ryzen CPU performing almost the same as the Intel CPU, aside from a couple of games where the game needs to understand Ryzen better, like Tomb Raider, even when you add all of the Avr FPS together the overall difference is still only 10%.

Correct, except with Ryzen you get a bunch more stuff for your money including extra cores that will be more and more useful over the next few years.

This is why when you test for CPU gaming performance you make the CPU work for it, if all the work is on the GPU then you would not see any difference in performance.
Aside from Tomb Raider there is no performance difference between these CPU's in Toms benchmarks.

If you make the CPU work for it, this happens.......

Forcing a CPU bottleneck when there wouldn't be one in reality might be interesting but it's not useful in determining which CPU is better for the task. Saying "there must be a winner" isn't an argument. I can't see your graph at work, by the way.

Why does that even matter ^^^^ ? because pepole upgrade thier GPU more than the CPU, GPU's get faster... so to keep up with future GPU performance CPU need to be faster.

This is based on the false premise that performance now will equal performance down the line. I've seen two videos showing aggregate benchmark scores showing, for example, that the FX-8xxx series outperforms the i5-2500K in modern gaming benchmarks, yet the i5-2500K was obviously much better at the time.

If all else was equal I'd agree that it'd be useful to know how the CPUs perform when they are more of a bottleneck during gaming, but considering there's a huge gap between 4c/4t and 6c/12t ($200 price point) and 4c/8t and 8c/16t ($300 price point), I do not believe for a second that the i7-7700K will be overall better for gaming over the next 3-5 years just because it's better now at 720p low settings.

Benchmarking processors with games is a joke period, it was only really 100% accurate in the days of software rendering. Once you bring other more important factors (GPU) into the equation it ceases to be a processor benchmark, the only semi-reliable way to benchmark processors with games today is to remove as much load as possible from the GPU by reducing resolution/details as much as possible.

Reliable in terms of repeatability but "is it actually useful" is the question.

I've argued countless times during the AMD FX years that most gaming benchmarks where AMD faired well were simply heavily GPU bottlenecked only to be told that "everyday usage is all that matters" but now that the shoe is on the foot AMD users are echoing what I said, only when you remove the GPU from the equation is the CPU able to stretch its legs.

Fair enough, I haven't had an AMD CPU in years so I don't know what kind of arguments people used to defend the FX-8xxx series, aside from "it's good for encoding".

 

[Curlyriff]

Benchmarking processors with games is a joke period, it was only really 100% accurate in the days of software rendering. Once you bring other more important factors (GPU) into the equation it ceases to be a processor benchmark, the only semi-reliable way to benchmark processors with games today is to remove as much load as possible from the GPU by reducing resolution/details as much as possible.

I've argued countless times during the AMD FX years that most gaming benchmarks where AMD faired well were simply heavily GPU bottlenecked only to be told that "everyday usage is all that matters" but now that the shoe is on the foot AMD users are echoing what I said, only when you remove the GPU from the equation is the CPU able to stretch its legs.

What are you on about. Almost everyone wants everyday usage still. It was only reviewers that decided otherwise and almost all of us said this is rubbish where are the 1080p/1440p/4k high & ultra setting benchmarks because all we want to know is how the CPU does in games compared to Intel if we brought it.

Not sure where you are getting all this "AMD users now want low 720p" stuff from? I don't remember seeing anyone advocate this tbh. Also it got disproved by a few reviewers that doing such test didn't actually prove anything in terms of future performance which was the reason "most" of the other reviewers were doing it apparently.

It isn't a joke period. I want to know what FPS I will get when I game at the settings I will use. I don't care about the theoretical difference between the two CPU's at something I will never do. The fact is that when actually gaming then both CPU's are fine with generally Ryzen now appearing to show it's strengths more in that it is able to almost keep up and in some cases surpass what Intel are doing, certainly lows and averages are great.

 

[Curlyriff]

In regards to the OP, I don't agree completely with what you are saying. I do understand what you are suggesting, but tbh I am not interested in theoretical performance provided by the evidence that is shown in your post.

To me doing the 720p low does not help me select a CPU. Why would I worry that I have 10% performance in games I play now if I can get 10% performance increase at actual game settings if that is what I am wanting. My biggest part though is I want to know what the CPU can do with games at 1440p/4K high/ultra settings and to see what the min/average/max and frame pacing is. This is what I will notice when gaming.

If I am in a much tighter grouped FPS range with Ryzen (which appears to be the case) then I would want this over an Intel CPU that may give bigger peak FPS but then has massive drops and frame pacing issues because of the 100% load which still happens at high resolution just of course it isn't constant like the 720p low would be.

That is where testing should be. Ideally to me you want to test the spectrum of CPU's in question at 1080/1440/4k with medium/high/ultra settings all showing frame pacing as well as the low/average/high and percentile frames.

Then that test needs to be run with half a dozen GPU's from low/medium/high from AMD & Nvidia to see how what difference is there. This then should also be run with different RAM speeds & timings accordingly to also show any differences there between the vendors.

To do that though would take weeks, game runs should be 15-30 mins each rather than a 30-45 seconds or even the 5 mins that most reviewers do. in-built game benchmarks should be ignored most the time as we have seen so many times now how they don't represent actual gaming performance whilst playing.

This should also be done over a number of games over the last few years, pref a dozen or so from a variety of different studios because then it can also highlight which developer may suit vendor also.

There is so much work there though that you would have to work doing it full time and would be probably a month behind everyone else and that would not happen as the 'professional' reviewers get paid on clicks and being that late to the party would mean dramatic reduction in traffic to their site and so revenue down. Better to get out the rubbish we have now for them than what we would really want to see.

This as well wont change because it would require every single reviewer signing up to do the same in depth methodology. And the only way to get that started would be for AMD to write a detailed document in doing so and not supply unless that review principle is followed. With that though the only variation different reviews would provide is different games or the same games to compare what different system setups are getting on average.

Not sure there really is a solution to reviews with the amount of information we would like to see prior to buying.

 

[Curlyriff]

How many of us when gaming just run a game on our pc when playing like the reviews? Most people i know also have email,chat programs chrome etc running which when you have more cores/threads means you maintain those ingame fps better then a cpu with less cores thats already fully loaded

I almost never have anything open at all on PC whilst gaming. I mainly play Rainbow Six Siege mainly and so no reason to have anything else open. If I am playing any action/fps game I will have everything closed. If I am playing something like Rome Total War then I may have Spotify open and that is it. I actually don't have anyone in my group that has more than either a music program and/or chat service of some sort open.

 

[mmj_uk]

What are you on about. Almost everyone wants everyday usage still. It was only reviewers that decided otherwise and almost all of us said this is rubbish where are the 1080p/1440p/4k high & ultra setting benchmarks because all we want to know is how the CPU does in games compared to Intel if we brought it.

Not sure where you are getting all this "AMD users now want low 720p" stuff from? I don't remember seeing anyone advocate this tbh. Also it got disproved by a few reviewers that doing such test didn't actually prove anything in terms of future performance which was the reason "most" of the other reviewers were doing it apparently.

It isn't a joke period. I want to know what FPS I will get when I game at the settings I will use. I don't care about the theoretical difference between the two CPU's at something I will never do. The fact is that when actually gaming then both CPU's are fine with generally Ryzen now appearing to show it's strengths more in that it is able to almost keep up and in some cases surpass what Intel are doing, certainly lows and averages are great.

If everyday usage in games is all that matters then CPU performance is largely irrelevant, just buy any decent quad core because GPU bottleneck will hamper anything better. Game benchmarks tell you very little about how a CPU performs unless you minimise the single most important factor in game performance (GPU).

Benchmarking a CPU using games at high resolutions (GPU bottlenecked) is like benchmarking a 1080Ti with an AMD Zacate processor constraining it, you don't get to see what it's really capable of.

 

[Curlyriff]

If everyday usage in games is all that matters then CPU performance is largely irrelevant, just buy any decent quad core. Game benchmarks tell you very little about how a CPU performs unless you minimise the single most important factor in game performance (GPU).

No because they do clearly have an affect otherwise we would see the same FPS regardless of the CPU. It has been shown that with the minimum with Ryzen already.

It still tells me exactly what I want. That I would on average get better performance on Ryzen for the games I want with much better low and average FPS.

Suggesting that it doesn't is obscene as otherwise there woudl be no argument and everyone should be on Ryzen as it's the cheaper option then.

 

[DarkHorizon472]

The benefits of Ryzen are often just the general benefits of having more than 4 cores. More now than in the past quad cores are being pushed to their limits by CPU intensive games. This means if any relatively intensive windows processes start in the background while gaming FPS can dive or start to stutter. Having more cores stops this happening. To give an example my 1700 runs BF1 at about 30% CPU in game on average. This means there is plenty of head room for anything else that starts up.

 

[icey_haj]

AdoredTV takes a look at whether Tom's Hardware is biased towards Intel or not. Quite an interesting watch.

 

[karik666]

Another very good Adored video.

 

[humbug]

AdoredTV takes a look at whether Tom's Hardware is biased towards Intel or not. Quite an interesting watch.

Anyone doubting the arguments as to why Toms Hardware made the wrong conclusions needs to watch that ^^^^

Toms own Data contradicts them.

 

[GAC]

nice video. guess more cores the better, ho hum more monies to spend !!!!

 

[haszek]

If everyday usage in games is all that matters then

What else is there? (if you are considering gaming)

I accept testing maximum CPU performance with no bottleneck, but demand conclusion "for majority, CPU doesn't matter in gaming, you won't see difference (except high refresh rate players)". If reviewers say it's no good for gaming just because it's weaker in 200-300fps range...than they shouldn't be doing "gaming reviews/conclusions"

 

[humbug]

Exactly, the same people said the 2500K, even the i3 was better for gaming than the FX-8350 because when the CPU is the bottleneck the i3 and i5 were faster.

Now that the Ryzen 5 is faster than the i5 in todays games the argument is turned on its head, now they are saying CPU bottlenecks don't matter because everyone runs graphics settings and resolutions relative to the performance of their graphics cards.

So was the AMD crowed right in saying exactly the same thing back then, can they have been wrong at the time but still be wrong today when those who said they are wrong use those methods today?

The truth is the situation between AMD and Intel CPU's has been turned on its head, so the reasoning gets turned on its head to suit the same agenda.

Logic also goes out the window, the Intel CPU running at 100% to keep up (often not) with the AMD CPU running at 50% is the better CPU than the AMD one..... madness, it boggles the mind.

 

[Martini1991]

When the CPU is the bottleneck it's not a rule that a Ryzen is faster than an Intel, because Intel still has the absolute best core for core performance, which is why in some games an Intel i5/i7 is still faster than any Ryzen chip, but the reverse can also be true on those games that can utilize the more threads, and ergo higher grunt of the Ryzens.

I never went low resolution though. 1080p's what matters for me as that's reasonable enough to show a CPU bottleneck in games as they actually matter, I've never once went on about 300 FPS 720P results, and in todays market at 1080p Ryzen CPU's and Intel lock horns, to a point that buying AMD is no longer a compromise, but should become far more dominate over any Intel 4 core.

To me if you're doing what you argued against, then you're just as flawed.

 

[humbug]

When the CPU is the bottleneck it's not a rule that a Ryzen is faster than an Intel, because Intel still has the absolute best core for core performance, which is why in some games an Intel i5/i7 is still faster than any Ryzen chip, but the reverse can also be true on those games that can utilize the more threads, and ergo higher grunt of the Ryzens.

I never went low resolution though. 1080p's what matters for me as that's reasonable enough to show a CPU bottleneck in games as they actually matter, I've never once went on about 300 FPS 720P results, and in todays market at 1080p Ryzen CPU's and Intel lock horns, to a point that buying AMD is no longer a compromise, but should become far more dominate over any Intel 4 core.

To me if you're doing what you argued against, then you're just as flawed.

You and i are in agreement.

 

[hyperseven]

AdoredTV takes a look at whether Tom's Hardware is biased towards Intel or not. Quite an interesting watch.

Is AdoredTV AMD biased?

 

[humbug]

Is AdoredTV biased towards AMD?

If you watch the video you would see that he is defending Toms Hardware in a lot of ways and he agrees with some of their conclusions.

Some things he doesn't agree with Toms, he doesn't agree that the "Best CPU's" are a complete 'blue wash', he goes on to explain why he thinks that, whats more he uses Toms Hardwares own reviews and numbers to explain why the Pentium, i3 and 7700K are right but the i5's are not, when you watch all of it you soon realise Toms Hardware based their conclusion on the wrong numbers, for example they used the Ryzen stock gaming performance vs the 7600K 5Ghz OC performance to come up with their numbers, and if you take out just one game, the Ryzen 'nVidia driver' troubled Tomb Raider the Ryzen 5 results higher performance than the i5's in any case, that Toms are also ignoring their own previous article on the fact that the i5's are at maximum capacity with the GTX 1080... just as a lot of other reviewers have shown, some unwittingly.

So no.

 

[icey_haj]

I don't think he's biased towards any side tbh. If you look at his videos, he's quite detailed and breaks down his analysis quite clearly. I think the method that he uses is fair and clear to see. It just so happens that it's showing up some of the more well known tech sites.