Watch Dogs CPU benchmarks, i7 (apparently) optional

[mikk]

del

 

[jj109]

Because their benchmarks usually tally up with what most people see with similar hardware on dozens of other games? Far less worse than other sites that posted Beta - even Alpha - builds of BF4 (simply to be "first post!") that ended up wildly different to the finished game...

Please tell me how TechSpot isn't "first posting" by releasing benchmarks on old drivers.

 

[AtenRa]

Interesting, from Toms Review the FX8350 is on par with Ivy Core i5 3550 and that by using one of the fastest dGPUs today. Seams WD is using more than 4 threads as the 6-core 3960X pulls away even at 1080p Ultra settings.


http://www.tomshardware.com/reviews/watch-dogs-pc-performance,3833-8.html

 

[escrow4]

It is indeed CPU heavy. The non K 4770 + 780 Ti GHz I have still had dips at 1200p to mid 40's from a solid 60 with Vsync everything maxed including that "pc" setting. I even had a BSOD on stock non overclocked (aside from factory OC GPU) well cooled hardware (GPU hit 72 Celsius, CPU less than 50). So its heavy and unoptimized and it doesn't look that pretty. FUN!

 

[erunion]

http://www.tomshardware.com/reviews/watch-dogs-pc-performance,3833-8.html

What a bizarre selection of CPUs!

 

[jj109]

What a bizarre selection of CPUs!

Yeah, you would have thought that doing all latest generation or something would be expected for a major site like THG.

I think deferred context rendering is rearing its head in PCGH and THG benches.

 

[AtenRa]

What a bizarre selection of CPUs!

I will agree they used a small selection of CPUs but bizarre ? why ??

 

[SPBHM]

even the 3770K is going as low as 22FPS with the Radeon, maybe it's time to change the title

it looks more like i5/i7 at 4.5GHz required ( if you want max details and always over 30 FPS with a Radeon specially).

 

[crashtech]

I will agree they used a small selection of CPUs but bizarre ? why ??

At least they should include a Haswell i3 if they are going to include the 3220.

 

[RussianSensation]

even the 3770K is going as low as 22FPS with the Radeon, maybe it's time to change the title

So at Tom's, Radeon R9 290X drops to 22 fps but at HardOCP it beats 780Ti? How can Tom's review even be taken seriously when they used FXAA and Medium textures and still got performance to drop to 22 fps at 1080P?

When reputable sites show 290X performing much closer to 780Ti than Tom's, one has to question their methodology and to be honest theirs leaves a serious question mark for me.

Another question to ask if it this game drops to 22 fps on a 3770K and a high end Radeon at 1080p but doesn't look better than Crysis 3 or Metro LL? What conclusion can be made? No excuses really for that level of optimization or lack thereof. The CPUs inside PS4/XB1 are crap compared to an i7 3770k, even at stock.

 

[SPBHM]

So at Tom's, Radeon R9 290X drops to 22 fps but at HardOCP it beats 780Ti? How can Tom's review even be taken seriously when they used FXAA and Medium textures and still got performance to drop to 22 fps at 1080P?

When reputable sites show 290X performing much closer to 780Ti than Tom's, one has to question their methodology and to be honest theirs leaves a serious question mark for me.

Another question to ask if it this game drops to 22 fps on a 3770K and a high end Radeon at 1080p but doesn't look better than Crysis 3 or Metro LL? What conclusion can be made? No excuses really for that level of optimization or lack thereof. The CPUs inside PS4/XB1 are crap compared to an i7 3770k, even at stock.

I'm talking about the pcgameshardware graphic comparing CPU scaling from nvidia and amd, nothing to do with tomshardware,

the game is running poorly considering how it looks sure, I guess DX12 is really needed, but also AMD software seems to be a lot less efficient, requiring a lot more CPU performance compared to nvidia, as you can see 3.4 vs 4.6GHz with both, also as usual nvidia gains a lot more with more CPU threads.

 

[Makaveli]

Require an i7 is not vague.

It certainly is considering there are 4 different generations of i7s floating around in the market.

 

[escrow4]

Why test at 720p? I mean seriously, this type of game will murder most CPUs at that res. Should have 1080/1200p minimum . . . . .

 

[toyota]

Why test at 720p? I mean seriously, this type of game will murder most CPUs at that res. Should have 1080/1200p minimum . . . . .

this should really not have to be explained but they test at 1280 to see if there are any real differences. if they tested at 1920 then you become more gpu limited and you dont know what performance you can expect if you turn down some graphics settings or run a faster gpu setup then whats in the review.

 

[erunion]

I will agree they used a small selection of CPUs but bizarre ? why ??

No mainstream i7, but they included an EE.
The 3 and 4 module chips are vishera, but the 2 module is BD.
A midlevel i5 was used, not the more common 3570.
The IVB parts are over 2 years old.

Clearly they weren't considering what would be most useful to readers considering a new CPU to play Watch Dogs on when they made their selection.

 

[Bubbleawsome]

I hope for driver optimizations, but it blows my mind that even my brand new i5 4670k couldn't max it at launch. I shouldn't need to rely on overclocking on a brand new enthusiast CPU. Even the extra 0.4Ghz I got on 1 core and the 0.7Ghz I got on 4 cores doesn't look like it would push my 75hz monitor. :|

 

[crashtech]

I hope for driver optimizations, but it blows my mind that even my brand new i5 4670k couldn't max it at launch. I shouldn't need to rely on overclocking on a brand new enthusiast CPU. Even the extra 0.4Ghz I got on 1 core and the 0.7Ghz I got on 4 cores doesn't look like it would push my 75hz monitor. :|

That's true for the 4960X as well, and that is a good thing! We want games that push the envelope. We should have games that need the highest-end card or SLI to play at Ultra settings at 60+ fps.

 

[Bubbleawsome]

That's true for the 4960X as well, and that is a good thing! We want games that push the envelope. We should have games that need the highest-end card or SLI to play at Ultra settings at 60+ fps.

While I agree, that only works if it looks the part. I'm sure I could code a program that uses 15 cores to just display a pixel of a random color. Minecraft pushes my 4670k to its limits when modded, and my GTX 770 well past its with shaders, but if vanilla did that would be broken. WDs isn't pushing limits, it's handicapped.

 

[Rakehellion]

It certainly is considering there are 4 different generations of i7s floating around in the market.

If a game runs on a slow i7 then it will run on a fast i5. Therefore it does not require an i7. That isn't ambiguous.

 

[TrantaLocked]

What confuses me is how the 4670K is matching the FX-9590 OC 5GHz in Watch Dogs. That just doesn't seem right, especially when the game was designed around AMD 8-core APUs in the consoles. I wonder if they were biased to cater to Intel-specific instructions for the PC version.

 

[davie jambo]

What confuses me is how the 4670K is matching the FX-9590 OC 5GHz in Watch Dogs. That just doesn't seem right, especially when the game was designed around AMD 8-core APUs in the consoles. I wonder if they were biased to cater to Intel-specific instructions for the PC version.

does not matter what cpu you use , the game runs like [garbage]

I've tried it fx8350 , i7 3770k and g850 (with 1x7970 , 2x7970 and 1x7850 cards)

does not run well on any of them

No profanity in the tech forums, please
-ViRGE

 

[SPBHM]

What confuses me is how the 4670K is matching the FX-9590 OC 5GHz in Watch Dogs. That just doesn't seem right, especially when the game was designed around AMD 8-core APUs in the consoles. I wonder if they were biased to cater to Intel-specific instructions for the PC version.

consoles run custom OS and "API/driver", PCs are running Windows, DX11, Forceware/Catalyst.

the consoles CPU is a totally different architecture compared to the 9590 (it's a lot slower) and as far as I know on the PS4 games only have access to 6 cores (1.6GHz Jaguar)

I think the PC is just paying the price for lack of optimization and Windows/DX11.

 

[Denithor]

Interesting, from Toms Review the FX8350 is on par with Ivy Core i5 3550 and that by using one of the fastest dGPUs today. Seams WD is using more than 4 threads as the 6-core 3960X pulls away even at 1080p Ultra settings.


http://www.tomshardware.com/reviews/watch-dogs-pc-performance,3833-8.html

Hmm.

All three Intel chips have a base speed of 3.3GHz. The top model is 6C/12T, middle is 4/4 and lowest is 2/4, making this pretty close to a comparison of core & thread counts.

Going from 2/4 to 4/4 nets a 54% increase in minimum fps.
Going from 4/4 to 6/12 nets a 38% increase in minimum fps.

So, my take from this is that the game is relatively heavily multi-threaded but even so there is a point of diminishing returns (performance plateaus off). Which also explains how well the AMD chips manage to do (the FX-8350 managing a small win over the i5-3550).

 

[sxr7171]

Long story short, your CPU matters if:

1. You are using SLI or Crossfire
2. You are running 120fps at reduced settings

 

[sxr7171]

It is indeed CPU heavy. The non K 4770 + 780 Ti GHz I have still had dips at 1200p to mid 40's from a solid 60 with Vsync everything maxed including that "pc" setting. I even had a BSOD on stock non overclocked (aside from factory OC GPU) well cooled hardware (GPU hit 72 Celsius, CPU less than 50). So its heavy and unoptimized and it doesn't look that pretty. FUN!

Watch dogs and Ubisoft are both a joke. That game runs so badly for no apparently reason. But are you sure it's CPU? Most people would say it's some kind of memory handling issue. I did the -disablepagefilecheck thing and while it improved things it still stutters at will.