Ryzen 5 1600 vs i7-7800X: Ryzen 1080p Gaming Issues Explained; 7700k still King

[Sauron]

11 minutes ago, mr moose said:

Again, price has nothing to do with it.

Not everyone buys the CPU solely for gaming, I did point that out.

Not everyone will buy a 1080ti, but then if they tested the 7700k or 1600 with rx 460 you'd have no idea if poor results were caused by GPU or CPU.

Everyone needs a reference.  At the moment the 7700K seems to be the defacto datum everything is measured against.  Whether you like it or not.

 

All that aside it appears you are the only person who has a problem with this.

If you're not buying a cpu solely for gaming then chances are you don't care about gaming performance as much as you care about whatever else it is that you're doing. And saying price doesn't matter, in any hardware comparison, doesn't make any sense.

 

I have a problem with the constant hammering on about gaming performance metrics which perpetrate the idea that that's all consumers care about. As for being the only one who has a problem with it, I remember being on the other side of a similar argument a few months ago when someone benchmarked a ryzen 1800x against a 7700k in 720p gaming. In that case I was ok with it because it was the first test with ryzen on the field, but since we're 6 months in and people are still running the same "baseline" test with cpus that are just lower core variants of the same thing (and which will perform almost exactly the same clock for clock in games) I'm kind of wondering if they even know you can do other things with a computer than playing games.

[tom_w141]

15 hours ago, VagabondWraith said:

-Snip regarding DF benchmarks-

Only 1 number? Assuming that's average, no 1% or .1% lows? That's some amateur hour shit.

Also 7900X at a lower clock speed consistently beating the 7700k/7740X What?  It's winning in games that don't scale beyond 4 cores. Something really wrong with that testing they need to revisit imo...

 

Off topic:

Though seeing numbers like this makes me happier I didn't wait for X299. Biggest difference is 21% at a frame rate where the difference is near impossible to differentiate and in a game that's an outlier for Ryzen (outliers should never be taken when creating an avg but still worth showing).

[mr moose]

1 minute ago, Sauron said:

If you're not buying a cpu solely for gaming then chances are you don't care about gaming performance as much as you care about whatever else it is that you're doing. And saying price doesn't matter, in any hardware comparison, doesn't make any sense.

 

I have a problem with the constant hammering on about gaming performance metrics which perpetrate the idea that that's all consumers care about. As for being the only one who has a problem with it, I remember being on the other side of a similar argument a few months ago when someone benchmarked a ryzen 1800x against a 7700k in 720p gaming. In that case I was ok with it because it was the first test with ryzen on the field, but since we're 6 months in and people are still running the same "baseline" test with cpus that are just lower core variants of the same thing (and which will perform almost exactly the same clock for clock in games) I'm kind of wondering if they even know you can do other things with a computer than playing games.

 

That's just an assumption, many people do both, they game and work.   Some (I would argue many) still only use a 1080 monitor. Not sure why that's hard to accept.

You seem to be stuck on the concept that price somehow effects performance. It doesn't, you can have two products that perform almost identical but one is twice as much,  should they not be compared?  of course they should. Only the consumer decides if the price metric is valid, not the reviewer.   The reviewer should test everything that could be reasonably expected to be wanted buy his audience.

 

Since ryzens launch,  it has given many home/small time builders a chance to look at new products for multiple tasks,But it's pretty bloody hard to work out how good ryzen is for those tasks if people like you insist they not be compared on grounds that one is a $120 more. 

[PCGuy_5960]

8 minutes ago, tom_w141 said:

Only 1 number? Assuming that's average, no 1% or .1% lows? That's some amateur hour shit.

Watch the video review. They have frametime charts

9 minutes ago, tom_w141 said:

Also 7900X at a lower clock speed consistently beating the 7700k/7740X What?  It's winning in games that don't scale beyond 4 cores. Something really wrong with that testing they need to revisit imo...

Most of the games that they tested scale beyond 4 cores and don't forget that Turbo Boost Max 3.0 boosts some cores to 4.7GHz

[tom_w141]

14 minutes ago, PCGuy_5960 said:

Most of the games that they tested scale beyond 4 cores and don't forget that Turbo Boost Max 3.0 boosts some cores to 4.7GHz

Apart from ashes of the benchmark I don't think any of those games scale particularly well? 7900X should not beat the 7740X/7700K based on historic data.

 

Spoiler

Spoiler

Excuse the age, why the fuck we still bench a game this old is beyond me  

Spoiler

 

[porina]

I'm losing track with who's arguing what now...

 

From my perspective, there are two interesting scenarios:

Similar configuration e.g. 1600(X) vs 7800X - this tells us about the architecture differences

Similar cost e.g. 1700(X) vs 7700k (vs 7800X at a stretch), or 1600(X) vs 7600k - this is the value question

Both have value, in their different ways.

 

I've been looking at going high fps 1080p for a while, and recently got a 7800X for such, before I heard/saw about it doing rather less well than 7700k. I admit I have some cash to throw around. Past testing showed even my OC 6600k destroys my OC 1700 in this use case, but it was still not enough. Dropping a 7700k in place of a 6700k didn't seem like enough of a difference, hence looking at 7800X as a balance between some extra cores, but hopefully still good clock potential. Maybe I should look at 7700k again... Or I could say stuff high fps 1080p, and get a 1440p g-sync or something. We certainly live in complicated times.

[XenosTech]

1 hour ago, Sauron said:

 

I still consider it a useless comparison given the use case and prices on the field. I don't know who buys a gtx 1080ti and a 7700k for 1080p, and I'm pretty sure it's not the overwhelming majority you seem to think it is, but regardless they have no use for this information - everyone already knew how this comparison would stack up. My point has never been that people don't play at 1080p, but rather that cpus haven't really mattered much in games for a while now. Yes, a 7700k may give you some 10-15% performance improvement on average at high framerates over a lower clocked i5 or ryzen 5, but for the most part it's just down to the graphics card - and the lower you go witht he gpu, the less the cpu matters.

What do you mean cpu's don't matter much in gaming ?  I know at 4k  they don't matter but at res below 1080p they matter a lot, 1080p they still matter and at 1440p they matter because if you pair a pentium with any above a 1060 you'll have one the worst gaming experiences. Cpu's don't matter at only at high res gaming where the GPU is the limiting factor

[PCGuy_5960]

5 minutes ago, tom_w141 said:

Apart from ashes of the benchmark I don't think any of those games scale particularly well? 7900X should not beat the 7740X/7700K based on historic data.

Crysis scales very well with more cores. Ryzen is able to beat the 7700K in this game

 

Witcher is GPU intensive so more cores don't make a huge difference.

 

Rise of the Tomb Raider, It depends on the game level AFAIK.

[tom_w141]

5 minutes ago, PCGuy_5960 said:

Crysis scales very well with more cores.

Doesn't seem to scale in the benchmark from when the game was current. Even if it did... A 30 fps disparity?? You don't think that's suspect?

 

5 minutes ago, PCGuy_5960 said:

Witcher is GPU intensive so more cores don't make a huge difference.

I agree so why is a slower 7900X beating a faster 7740X by 14 fps?

 

5 minutes ago, PCGuy_5960 said:

Rise of the Tomb Raider, It depends on the game level AFAIK.

Maybe but they would have used the game's benchmark for consistency. In the above chart it doesn't scale with cores. Again 16 fps more on the slower 7900X??

 

In Conclusion:

 

THESE NUMBERS ARE WHACK  

 

EDIT: Something I didn't spot till now.. Look at these sick gains when they OC! 1 fps, 2 fps, 9 fps, 1 fps, 2 fps, 1 fps,4 fps WOOOOOOOOW

[Taf the Ghost]

@tom_w141

 

Actually, looking at the numbers more closely, unless it's a really good game for taking more cores, I think the issue is actually a GPU bottleneck that the 7700k hits more easily.

[tom_w141]

23 minutes ago, Taf the Ghost said:

@tom_w141

 

Actually, looking at the numbers more closely, unless it's a really good game for taking more cores, I think the issue is actually a GPU bottleneck that the 7700k hits more easily.

My point is DF's numbers are fucked because with the same GPU (Titan X Pascal) it shows a slower clocked 7900X beating a higher clocked 7740X by quite a large margin in games that don't scale well with threads (ashes of the benchmark excluded)

[VagabondWraith]

3 hours ago, tom_w141 said:

Only 1 number? Assuming that's average, no 1% or .1% lows? That's some amateur hour shit.

Also 7900X at a lower clock speed consistently beating the 7700k/7740X What?  It's winning in games that don't scale beyond 4 cores. Something really wrong with that testing they need to revisit imo...

 

Off topic:

Though seeing numbers like this makes me happier I didn't wait for X299. Biggest difference is 21% at a frame rate where the difference is near impossible to differentiate and in a game that's an outlier for Ryzen (outliers should never be taken when creating an avg but still worth showing).

Quote

In terms of gaming, we've since tightened up our testing procedures to ensure that the CPU takes point as much as possible as the bottleneck in performance. We aim to concentrate on the most CPU-intensive areas of our titles, while offering a repeatable test that we can roll out across all processors. Our results aren't designed to show performance in a usual gaming scenario (where GPU is usually the defining factor), more to show relative maximum theoretical performance using a cross-section of challenging game engines. 

Actually isolating areas of gameplay that constitute a thorough CPU work-out isn't easy - and there's still the sense that with high-end processors like this, we may still be GPU-bound in some areas. There are plenty of scenarios where we can get the best out of all of these powerful processors though. We've relocated our Crysis 3 GPU bench area to a section of the Welcome to the Jungle stage that can challenge even a ten-core i7, and we've binned off Rise of the Tomb Raider's inbuilt GPU-centric benchmark for actual gameplay that pushes CPU hard in the notorious Geothermal Valley.

We do keep our gallop through Novigrad City in The Witcher 3 though - just like Crysis 3, this area of gameplay can see a mainstream i5 hit 100 per cent utilisation across all cores. Far Cry Primal remains an excellent workout for single-core performance in a modern engine, while Assassin's Creed Unity and The Division are both many-core aware, but tend to max out on a standard quad-core i7. Meanwhile, Oxide Games' Ashes of the Singularity CPU stress test does exactly what it says on the tin - however many cores your processor has, it'll do its level best to make good use of them.

 

On top of that, performance minimums are all above the crucial 60 frames per second. And this is all with no overclocking

 

[tom_w141]

10 minutes ago, VagabondWraith said:

while Assassin's Creed Unity and The Division are both many-core aware, but tend to max out on a standard quad-core i7. Meanwhile, Oxide Games' Ashes of the Singularity CPU stress test does exactly what it says on the tin - however many cores your processor has, it'll do its level best to make good use of them.

Rest is irrelevant but thanks for this part it confirms my point. The above games where the slower 10 core wins should max out at 4 cores 8 threads. SO HOW ON EARTH DOES A SLOWER 10 CORE BEAT A FASTER QUAD CORE?! THOSE NUMBERS ARE WRONG SOMEWHERE.

 

Of all the games they tested the only 1 where the slower 7900X should win is ashes.

 

EDIT: Crysis is the biggest disparity beats it by 30 fps in game only using 4 cores. 14 in Witcher and 16 in Tomb Raider are all also games that max out on a quad core yet get beaten by a slower 10 core that should have 6 of its cores parked.

 

(If these games were truly multithreaded as this review would indicate then an R7 would crush a 7700K/7740X) That's how we know this is BS.

[VagabondWraith]

Just now, tom_w141 said:

Rest is irrelevant but thanks for this part it confirms my point. The above games where the slower 10 core wins should max out at 4 cores 8 threads. SO HOW ON EARTH DOES A SLOWER 10 CORE BEAT A FASTER QUAD CORE?! THOSE NUMBERS ARE WRONG SOMEWHERE.

 

Of all the games they tested the only 1 where the slower 7900X should win is ashes.

Why you yelling at me?  I didn't do the test. If you have a problem, take it up with Digital Foundry. I'm just showing you their findings. You choose to believe it or not, that's not my problem. All I'm saying is, Skylake-X can game, and do it damn well.

[tom_w141]

Just now, VagabondWraith said:

Why you yelling at me?

wasn't yelling sorry just making the obvious error stand out. Its not possible (further reasoning edited in above)  

 

1 minute ago, VagabondWraith said:

You choose to believe it or not

When 1 source's results contradict all others its not a matter of choosing whether to believe.

 

2 minutes ago, VagabondWraith said:

All I'm saying is, Skylake-X can game, and do it damn well

Agreed, but as this thread says it still can't beat the mainstream offering (which is sad imo, my idea of enthusiast = best in every category), also until more people test the 7800X I'm not ruling out an issue there. The others seem ok in all other testing but still beat by the 7700K. DF should sell those magical chips though that can somehow crank out more fps with the same IPC and lower clock speeds! Witchcraft  

[Drak3]

1 hour ago, tom_w141 said:

Rest is irrelevant but thanks for this part it confirms my point. The above games where the slower 10 core wins should max out at 4 cores 8 threads. SO HOW ON EARTH DOES A SLOWER 10 CORE BEAT A FASTER QUAD CORE?! THOSE NUMBERS ARE WRONG SOMEWHERE.

 

Of all the games they tested the only 1 where the slower 7900X should win is ashes.

 

EDIT: Crysis is the biggest disparity beats it by 30 fps in game only using 4 cores. 14 in Witcher and 16 in Tomb Raider are all also games that max out on a quad core yet get beaten by a slower 10 core that should have 6 of its cores parked.

 

(If these games were truly multithreaded as this review would indicate then an R7 would crush a 7700K/7740X) That's how we know this is BS.

Hey sherlock, background processes are a constant thing happening, and there is benefit from having more threads than a game can max out, to prevent conflict between those processes and the game trying to run on the same logical cores.

It'd make perfect sense if we knew what exactly was running, every little bit of it.

[tom_w141]

2 minutes ago, Drak3 said:

Hey sherlock, background processes are a constant thing happening, and there is benefit from having more threads than a game can max out, to prevent conflict between those processes and the game trying to run on the same logical cores.

It'd make perfect sense if we knew what exactly was running, every little bit of it.

Anyone benchmarking on anything less than a clean install with no other programs is a fool and should have their numbers ignored anyway.

[Drak3]

Just now, tom_w141 said:

Anyone benchmarking on anything less than a clean install with no other programs is a fool and should have their numbers ignored anyway.

Operating systems also have their own background processes.

Also, should a reviewer be consistant in their testing methodology, their numbers should not br ignored either. Numbers closer to real world usage are valuable to.

[tom_w141]

13 minutes ago, Drak3 said:

Operating systems also have their own background processes.

That utilise 1-2% cpu...

 

13 minutes ago, Drak3 said:

Numbers closer to real world usage are valuable to.

If a reviewer clogged their PC with everyday use processes it would close the gap or make the Ryzen and i7 level in performance. In that instance the reviewer would be crucified, but if its Intel beating Intel thats fine and its nice they are testing "realistic conditions" mmm k  

 

Back to the original point: trying to say windows background processes would cost as much as 30 fps is absolute bull shit.

[Drak3]

18 minutes ago, tom_w141 said:

That utilise 1-2% cpu...

Not always, and that is extremely dependant on how your task manager of choice reports it.

[tom_w141]

14 minutes ago, Drak3 said:

Not always, and that is extremely dependant on how your task manager of choice reports it.

Splitting hairs here. Numbers look incorrect. 7900X when slower than the 7740X shouldn't have won in any of those titles except ashes. Definitely not by 15-30 fps. Something. Is. Wrong.

[Drak3]

45 minutes ago, tom_w141 said:

Splitting hairs here. Numbers look incorrect. 7900X when slower than the 7740X shouldn't have won in any of those titles except ashes. Definitely not by 15-30 fps. Something. Is. Wrong.

Cache configuration can also play a role. The Broadwell 5775C could match the 6700K and 4790K in games while being clocked lower due to having an L4 cache: it's 128MB DRAM.

[tom_w141]

6 minutes ago, Drak3 said:

Cache configuration can also play a role. The Broadwell 5775C could match the 6700K and 4790K in games while being clocked lower due to having an L4 cache: it's 128MB DRAM.

Or the DF results are garbage. This is how it should look

 

Spoiler

Spoiler

 

[Demonking]

1 hour ago, tom_w141 said:

Anyone benchmarking on anything less than a clean install with no other programs is a fool and should have their numbers ignored anyway.

Huh ? wouldn't that be representing real world performance normal gamers don't play on PC without there apps they need/use.

[tom_w141]

1 minute ago, Demonking said:

Huh ? wouldn't that be representing real world performance normal gamers don't play on PC without there apps they need/use.

Clean slate is the fairest comparison and removes all other factors. At the end of the day the job is to compare cpu vs cpu removing as many other variables as possible. Which is why an overkill gpu is used and 1080p to focus on the gpu and minimise gpu bottleneck. The reviewer can't anticipate everyone's "normal" programs so it is best to bench without them. For example 1 person's every day programs might just be steam, another has steam teamspeak afterburner OBS etc... Its a more accurate comparison to use a clean build. I doubt DF tested with an everyday use PC. Fresh = Fastest and thats what they want.