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

[VagabondWraith]

I may conduct my own tests with some of these games by disabling 2 cores and running these benchmarks and report back.

[Valentyn]

1 minute ago, PCGuy_5960 said:

I am pretty sure that this is just a screenshot of the benchmark run, not the actual benchmark

With a 7900X though. Need I remind you that Hardware Unboxed also got worse performance than the 7700K with the 7740X? (Proven wrong by GN's benchmarks)

I really doubt it, I believe that the AsRock board doesn't work well with the lower end X299 CPUs (The 7800X and the 7740X) 

Mind linking those please, as far as I recall his 7900X results matched that of OC3D and PcPer. PcPer having the worst 7900X results of all, and used the Asus Prime X299 Deluxe.

 

Infact TechSpot got better results on the Tachi with the 7900X than PcPer did with the Asus board on some games tests.

https://www.techspot.com/review/1433-intel-core-i9-core-i7-skylake-x/page3.html

 

 

https://www.pcper.com/reviews/Processors/Intel-Core-i9-7900X-10-core-Skylake-X-Processor-Review/1080p-Gaming-Performance-a

 

[Valentyn]

2 minutes ago, VagabondWraith said:

I may conduct my own tests with some of these games by disabling 2 cores and running these benchmarks and report back.

Also have to disable some cache and cripple your IMC a bit. The 7800X is worse in both those regards than the 7820X.

[RotoCoreOne]

6 cores still going to 400$? Bleh. Rather odd that at 4.7ghz the 7800X barely edges out the 1600@4.0Ghz. Or is it just me...

[PCGuy_5960]

Just now, Valentyn said:

Mind linking those please, as far as I recall his 7900X results matched that of OC3D and PcPer. PcPer having the worst 7900X results of all, and used the Asus Prime X299 Deluxe.

 

Infact TechSpot got better results on the Tachi with the 7900X than PcPer did with the Asus board on some games tests.

I meant that tomshardware used the 7900X in the tests that you linked (in which the Taichi was on par with the Deluxe) 

 

TechSpot's 7740X results were very bad and GN (using an ASUS board) got much better results, this is why I believe that this is an AsRock issue. Their boards may not be very well optimized for the lower end X299 CPUs

[Valentyn]

6 minutes ago, PCGuy_5960 said:

I meant that tomshardware used the 7900X in the tests that you linked (in which the Taichi was on par with the Deluxe) 

 

TechSpot's 7740X results were very bad and GN (using an ASUS board) got much better results, this is why I believe that this is an AsRock issue. Their boards may not be very well optimized for the lower end X299 CPUs

Could be, but what is there to optimise really?

What we need is to see GN and other also run 7800X tests to compare. Can't really use a sample size of two with Taichi and 7800X to declare it's the board.

 

Anandtech did a fantastic in-depth review of Skylake X, including the 7800X; but no gaming tests at all.

 

Vortez did some tests, but they used a bloody RX 480..... and their 7700K was doing terrible. The fuck were they doing?

 

Note they use the Gigabyte Auros X299-Gaming 3

 

[Valentyn]

6 minutes ago, RotoCoreOne said:

6 cores still going to 400$? Bleh. Rather odd that at 4.7ghz the 7800X barely edges out the 1600@4.0Ghz. Or is it just me...

If you want a 6 core with great IPC wait for Coffee Lake on Z370+ or Z380.
Else, if you're on a budget get the 1600.

[Taf the Ghost]

1 hour ago, VagabondWraith said:

My question is, why is he using a GTX 1050 in some of his benchmarks? Why is he capping his framerate? His testing looks flawed. Digital Foundry found that the 7800X often beats the 7740X.

 

 

 

Looks like he just used the wrong test footage for that spot.

16 minutes ago, goodtofufriday said:

Am I missing something? Isn't this a "No Duh kinda thing? i7 vs an R5?? Why are we wasting time comparing the two. 

There's a lot going on with these benchmarks. Nothing really new with the R5 1600 (best value) or 7700k (best outright gaming performance with a few rare exceptions for now), but the 7800X is new and a direct comparison with the Ryzen 5 1600 has value. At minimum, we know the 7800X isn't a better Gaming CPU, but it points to a whole lot of other aspects for things going forward. 

 

Though I'm hoping we're finally reaching the point where "remove the GPU bottleneck" is ever stated when testing CPUs. Specific aspects of CPU Architecture & GPU driver implementation are big issues with Gaming and that simply can't be worked around.

[RotoCoreOne]

Just now, Valentyn said:

If you want a 6 core with great IPC wait for Coffee Lake on Z370+ or Z380.
Else, if you're on a budget get the 1600.

Oh, I have a 1700X. I'm just lamenting the fact that it still costs so much from intel for not much better performance.

[PCGuy_5960]

1 minute ago, Valentyn said:

Could be, but what is there to optimise really?

Same as Ryzen. Because the launch was rushed, it is possible that AsRock only optimized for the 7900X and the 7820X (the 2 highest end CPUs)

[Valentyn]

2 minutes ago, PCGuy_5960 said:

Same as Ryzen. Because the launch was rushed, it is possible that AsRock only optimized for the 7900X and the 7820X (the 2 highest end CPUs)

Impossible to tell given we only have 2 reliable* reviews of the 7800X so far, and both were done on the Taichi.

 

The only other review I could find with gaming included is that Vortez one, and they seriously botched it with a horrible GPU bottleneck.

 

Either way though, my money is starting to burn me. I got an X299 system specced and ready to order. Just need Threadripper to show up!

If I go X299, I'll go for the Asus TUF Mark 1.

 

[Taf the Ghost]

Just now, PCGuy_5960 said:

Same as Ryzen. Because the launch was rushed, it is possible that AsRock only optimized for the 7900X and the 7820X (the 2 highest end CPUs)

Not impossible. Also, the Skylake-X CPUs scale in L3 cache as it goes up in SKU. It's very possible that starts to cause issues at the really extreme edge of the Performance scaling. When you're down at 5 ms per frame (200 FPS), very subtle timing/design issues start to crop up. 

3 minutes ago, Valentyn said:

Could be, but what is there to optimise really?

What we need is to see GN and other also run 7800X tests to compare. Can't really use a sample size of two with Taichi and 7800X to declare it's the board.

 

Anandtech did a fantastic in-depth review of Skylake X, including the 7800X; but no gaming tests at all.

 

Vortez did some tests, but they used a bloody RX 480..... and their 7700K was doing terrible. The fuck were they doing?

 

Note they use the Gigabyte Auros X299-Gaming 3

uArch matters. A lot more than we previously were willing to admit. Also, noticeably, it's an AMD GPU that isn't going to get that Nvidia bump from specific Optimizations. Stuff like that actually isn't unexpected. AMD is going to be better equipped to optimize drivers to their uArch.

[Taf the Ghost]

4 minutes ago, Valentyn said:

Impossible to tell given we only have 2 reliable* reviews of the 7800X so far, and both were done on the Taichi.

 

The only other review I could find with gaming included is that Vortez one, and they seriously botched it with a horrible GPU bottleneck.

 

Either way though, my money is starting to burn me. I got an X299 system specced and ready to order. Just need Threadripper to show up!

If I go X299, I'll go for the Asus TUF Mark 1.

 

AMD is going to have a lot of announcements out of SIGGRAPH. And isn't August 10th TR's expected launch?  

 

What's your Use Case?

[PCGuy_5960]

Just now, Taf the Ghost said:

Not impossible. Also, the Skylake-X CPUs scale in L3 cache as it goes up in SKU. It's very possible that starts to cause issues at the really extreme edge of the Performance scaling. When you're down at 5 ms per frame (200 FPS), very subtle timing/design issues start to crop up. 

Yes, but Salazar (Science Studio) also used the same AsRock Taichi board and got similar gaming performance. It's probably a board optimization issue, but more testing needs to be done to determine if that's true or not. Let's hope that GN reviews the 7800X

[TheOriginalHero]

35 minutes ago, goodtofufriday said:

Am I missing something? Isn't this a "No Duh kinda thing? i7 vs an R5?? Why are we wasting time comparing the two. 

The video isn't really about the r5 vs 7700k but more the r5 vs the 7800x. Intel's 6 core 12 thread cpu clocked at 4.7ghz is matching the gaming performance of the 1600 clocked at only 4ghz. Considering the huge price difference between the 2 platforms and the 700mhz and ipc advantage on intels side many people are baffled as to why the 7800x is performing so poorly. The i7 7700k was included as a reference point as many expected the gaming performance of the 7800x and the 7700k to be similar considering there's only a 200mhz difference.

[Valentyn]

6 minutes ago, Taf the Ghost said:

AMD is going to have a lot of announcements out of SIGGRAPH. And isn't August 10th TR's expected launch?  

 

What's your Use Case?

Video encoding, rendering, and gaming/recording/streaming.
So I would love some cores, but also want some decent gaming.

( Soon lots of VM and remote work as well )

 

8-12 cores would be my sweet spot. Main issue for gaming is I'm still an avid WoW player and single core performance is needed there. One of the main reasons I'm still considering X299; since the dualcore boosts are great.

 

Threadripper shows 4.0Ghz boost with XFR. Usually that's 2 cores, but since Threadripper has two modules, I do wonder if it'll boost 2 cores per modules then. Given 4 cores at 4.0Ghz.

If that's the case, and I can get the memory ar 3200Mhz + I'll happily snap up the 12 core version for my system.

It'll be the best all rounder then compared to the 7820X; which would be better in the gaming department, but suffer in everything else.

 

I must say the X399 Asus Zenith board looks fantastic, I just hope it's priced well. If the motherboards are silly prices it'll defeat AMD's platform cost advantage.

[Taf the Ghost]

2 minutes ago, PCGuy_5960 said:

Yes, but Salazar (Science Studio) also used the same AsRock Taichi board and got similar gaming performance. It's probably a board optimization issue, but more testing needs to be done to determine if that's true or not. Let's hope that GN reviews the 7800X

Maybe. I'm of the opinion that it's really down to the fact that Intel had sand-bagged so long on any larger changes that Nvidia was able to find ~15-20% more performance with Maxwell/Pascal due to some really clever optimizations of the Intel uArch. Once those no longer work, suddenly there is "regression". It also goes to the issues with Nvidia's DX12 implementation. But there still could be an issue with the Taichi. Wouldn't be the first launch board with odd problems.

[Taf the Ghost]

4 minutes ago, Valentyn said:

Video encoding, rendering, and gaming/recording/streaming.
So I would love some cores, but also want some decent gaming.

( Soon lots of VM and remote work as well )

 

8-12 cores would be my sweet spot. Main issue for gaming is I'm still an avid WoW player and single core performance is needed there. One of the main reasons I'm still considering X299; since the dualcore boosts are great.

 

Threadripper shows 4.0Ghz boost with XFR. Usually that's 2 cores, but since Threadripper has two modules, I do wonder if it'll boost 2 cores per modules then. Given 4 cores at 4.0Ghz.

If that's the case, and I can get the memory ar 3200Mhz + I'll happily snap up the 12 core version for my system.

It'll be the best all rounder then compared to the 7820X; which would be better in the gaming department, but suffer in everything else.

 

I must say the X399 Asus Zenith board looks fantastic, I just hope it's priced well. If the motherboards are silly prices it'll defeat AMD's platform cost advantage.

Ah, okay, I can see wanting a bunch of benchmarks. I might suggest Process Lasso with any TR-based build, because segmenting a game all onto 1 CCX does produce an advantage. I don't expect WoW to be an issue, as the GPU is almost assuredly the bottleneck there. (And we're talking, what, 350 fps vs 320 fps?  ) But if you're running multi-monitor, that makes life a lot wackier. Gaming + Multi-monitor setups is something that really needs more testing these days.

 

I'm in something of the same boat. No new computer this year (I've had to smack my hands about it already, haha), but I'm really interested in a 16c TR build next year. I have a use for a lot of cores but I don't need all of the speed. I don't AAA game and I have a use for multiple VMs + OSes. (And let me move from multiple computers to 1.) So I'm really curious where you decide to go as well.

[raphidy]

RIP just received my 1700. x370 prime pro this weekend.

[Valentyn]

1 minute ago, Taf the Ghost said:

Ah, okay, I can see wanting a bunch of benchmarks. I might suggest Process Lasso with any TR-based build, because segmenting a game all onto 1 CCX does produce an advantage. I don't expect WoW to be an issue, as the GPU is almost assuredly the bottleneck there. (And we're talking, what, 350 fps vs 320 fps?  ) But if you're running multi-monitor, that makes life a lot wackier. Gaming + Multi-monitor setups is something that really needs more testing these days.

 

I'm in something of the same boat. No new computer this year (I've had to smack my hands about it already, haha), but I'm really interested in a 16c TR build next year. I have a use for a lot of cores but I don't need all of the speed. I don't AAA game and I have a use for multiple VMs + OSes. (And let me move from multiple computers to 1.) So I'm really curious where you decide to go as well.

I wish we were talking about FPS like that.

The latest expansion increased all the fidelity, they redid decade old models and more.

At the moment in some raids I'll easily sit around the 40's with non maxed settings on the system in my signature. People with GTX 1080Ti's not able to max out the settings or they're in the low 30's at times.

 

Old video of my guild, but you can see the FPS in the top right under the minimap.

https://www.youtube.com/watch?v=KX6cL801Qos

The game's alpha particle effects, draw distance, ground clutter, environmental effects, and shadows are all on that single CPU core that has to deal with the rest of the game as well.
 

 

One day they might update it to DX12 or Vulka;, on OSX moving to Metal gave significant performance increases for WoW.

 

Also yes, gaming in Multi-monitor because I always have stuff to check up on, and manage. The extra cores help there as well. Easier to game and do more; just annoying with some games.

 

I'm hoping reviews and platform costs for Threadripper are solid. Especially if it has XFR boost 4 cores, as opposed to 2 on Ryzen. Would mean I won't have to even bother looking into overclocking ( much better for work ).

 

I'm also curious as to where I'll go as well. I'm the prime candidate for HEDT I think. Work and play on the same system; and since I'll be keeping it for 3+ years ( bar GPU ), I really want to make the correct choice.

[VagabondWraith]

40 minutes ago, Taf the Ghost said:

Not impossible. Also, the Skylake-X CPUs scale in L3 cache as it goes up in SKU. It's very possible that starts to cause issues at the really extreme edge of the Performance scaling. When you're down at 5 ms per frame (200 FPS), very subtle timing/design issues start to crop up. 

uArch matters. A lot more than we previously were willing to admit. Also, noticeably, it's an AMD GPU that isn't going to get that Nvidia bump from specific Optimizations. Stuff like that actually isn't unexpected. AMD is going to be better equipped to optimize drivers to their uArch.

That's what I noticed with Digital Foundries video; performance scaling up the tier. 7900X was sometimes 10 FPS ahead of the 7820X; whereas both were considerably ahead of the 7800X.

 

It shouldn't be a cache issue as all Sky-X CPU's have the same amount of L3 at 1.375 per core. Maybe it's because the 7800X has less total L3 than the other two, but then again, it still has more total L3 than the 7700K/7740X.

[Taf the Ghost]

3 minutes ago, VagabondWraith said:

That's what I noticed with Digital Foundries video; performance scaling up the tier. 7900X was sometimes 10 FPS ahead of the 7820X; whereas both were considerably ahead of the 7800X.

 

It shouldn't be a cache issue as all Sky-X CPU's have the same amount of L3 at 1.375 per core. Maybe it's because the 7800X has less total L3 than the other two, but then again, it still has more total L3 than the 7700K/7740X.

I've noticed too that the 7820X and 7900X seem to both scale better in gaming and, generally though not completely normalized, be better than the 7800X.  However, the 7700k tends to win just about everything still. It's something to do with the way the L3 cache is used. We're getting into nanosecond-level timing efficiencies where you'd need a lot of very careful testing to really sort out exactly what is going on. And most of that is still in the way the Nvidia driver does its processing, which only comes up with at least the 1080.

[Valentyn]

6 minutes ago, VagabondWraith said:

That's what I noticed with Digital Foundries video; performance scaling up the tier. 7900X was sometimes 10 FPS ahead of the 7820X; whereas both were considerably ahead of the 7800X.

 

It shouldn't be a cache issue as all Sky-X CPU's have the same amount of L3 at 1.375 per core. Maybe it's because the 7800X has less total L3 than the other two, but then again, it still has more total L3 than the 7700K/7740X.

The Kabylake parts also have the old Ring Bus system that just seems to "mesh" better with games.

The 7800X seems really odd; even if it's issueless, Coffee Lake is just around the corner. That'll give 6 cores 12 threads, and use the Ring Bus like Kabylake, along with having more L3 cache as well.

Add that on to a cheaper platform, and it just seems far better than the 7800X. 

 

The way I see it, if you're going X299 you go 8 core 10+ cores, or Kabylake X for pure performance.

[Taf the Ghost]

29 minutes ago, Valentyn said:

I wish we were talking about FPS like that.

The latest expansion increased all the fidelity, they redid decade old models and more.

At the moment in some raids I'll easily sit around the 40's with non maxed settings on the system in my signature. People with GTX 1080Ti's not able to max out the settings or they're in the low 30's at times.

 

Old video of my guild, but you can see the FPS in the top right under the minimap.

https://www.youtube.com/watch?v=KX6cL801Qos

The game's alpha particle effects, draw distance, ground clutter, environmental effects, and shadows are all on that single CPU core that has to deal with the rest of the game as well.
 

 

One day they might update it to DX12 or Vulka;, on OSX moving to Metal gave significant performance increases for WoW.

 

Also yes, gaming in Multi-monitor because I always have stuff to check up on, and manage. The extra cores help there as well. Easier to game and do more; just annoying with some games.

 

I'm hoping reviews and platform costs for Threadripper are solid. Especially if it has XFR boost 4 cores, as opposed to 2 on Ryzen. Would mean I won't have to even bother looking into overclocking ( much better for work ).

 

I'm also curious as to where I'll go as well. I'm the prime candidate for HEDT I think. Work and play on the same system; and since I'll be keeping it for 3+ years ( bar GPU ), I really want to make the correct choice.

Wow, that's really sad about... WoW.

[Valentyn]

6 minutes ago, Taf the Ghost said:

Wow, that's really said about... WoW.

Still a fantastic game, and we're enjoying it so much at the highest levels of play they have to offer. 

 

Shame it's still built on tech from 2001 ( modified Warcraft 3 engine ). The game is riddled with spaghetti code; they can't even increase the backpack size without inadvertently breaking something elsewhere in the game.

 

And they've tried for years now.

 

I love their art style and aesthetic, but my god is the game a chore to run at times. 

This is what happens if you have nearly 80 people in one spot duking it out with spell effects everywhere. At the time I had SLI 980Ti's as well.

 

FPS under the minimap in the top right corner as well; this was at 1440p.

 

 

The main solution is to try and shift more to the GPU, upping resolution and using effects that affect it more. Even then the engine loves a single core.

If I wasn't such a WoW-crackhead I'd have no issues deciding which to get. More cores ad cheaper; as long as they do well at 1440p minimum.