60 Minutes – CPU performance and Energy consumption in Gaming. [AtenRa]

[erunion]

That analogy would only be valid if people drove their corvettes in real life with S rated tires for the most part. AtenRa's testing represents how I think most gamers actually use their PC. It isn't isolating just the CPU if it's GPU limited, I get that, you are right. But I think most people actually game at mostly GPU limited settings.

This has been heavily discussed but I'm of the opinion that CPU bottlenecks are much worse than GPU bottlenecks in real life scenarios. I accept that many gamers are GPU limited, and that's a good thing.
A GPU bottleneck causing unusable FPS can be fixed easily by changing resolution/detail settings.
Also, people upgrade GPUs more often than CPUs. If a rig has sufficient CPU to be GPU limited that means it will be a good candidate for a GPU upgrade. When the new 14/16nm GPUs come out soon it will trigger an upgrade cycle and I predict we will see a lot of people will find themselves with GPU heavy/cpu light rigs.

In other words the maximum potential of CPUs does matter, even for people who don't buy top end GPUs.

 

[Denithor]

Civ:BE seems to be the least GPU bound of your games tested. I wonder how a similarly overclocked PII x6 and a 6-core Xeon would look in that test?

 

[AtenRa]

I was looking at the VC&G forums and this thread caught my eye, so i installed my FX8350 and started playing the game.

http://schedule.gdconf.com/session/...ng-for-multi-core-and-dx12-presented-by-intel

Codemasters present a post-mortem on their new rendering engine used for F1 2015 detailing how they balanced the apparently opposing goals optimizing for mainstream processor graphics, high end multi-core and DX12. The F1 2015 engine is Codemasters' first to target the eighth generation of consoles and PC's with a new engine architecture designed from scratch to distribute the games workload across many cores making it a great candidate for DX12 and utilise the processing power of high end PC's.

Very interesting, and this is on DX-11.

 

[AtenRa]

Civ:BE seems to be the least GPU bound of your games tested. I wonder how a similarly overclocked PII x6 and a 6-core Xeon would look in that test?

I got my self a nice Phenom II 1090T, so give me some time and stay tuned

 

[SlowSpyder]

I got my self a nice Phenom II 1090T, so give me some time and stay tuned

Mine ran happily for years at 4GHz in my AM2+ system. Decent CPU in it's day with a little staying power thanks to all those cores.

 

[Zor Prime]

Mine ran happily for years at 4GHz in my AM2+ system. Decent CPU in it's day with a little staying power thanks to all those cores.

I still run a 1045 at 3.5GHz and score 60FPS easy at 1080p in titles such as Witcher 3, Fallout 4, etc ... with ultra/high settings with a 960 4GB card.

I'm effectively gonna get to skip the entire Bulldozer series of processors until Zen thanks to Phenom II X6.

 

[crashtech]

Subscribed, wish we could see how a more modern quad does against the old 2500K.

 

[VirtualLarry]

I still run a 1045 at 3.5GHz and score 60FPS easy at 1080p in titles such as Witcher 3, Fallout 4, etc ... with ultra/high settings with a 960 4GB card.

Really??? Wow. Maybe I should pull out my 1045T (that I also had at 3.5Ghz-ish, at stock voltage even), and use it. Was thinking about that lately.

 

[frozentundra123456]

I still run a 1045 at 3.5GHz and score 60FPS easy at 1080p in titles such as Witcher 3, Fallout 4, etc ... with ultra/high settings with a 960 4GB card.

I'm effectively gonna get to skip the entire Bulldozer series of processors until Zen thanks to Phenom II X6.

Those seem like pretty ...optimistic numbers for a 960. Not sure how 60FPS could be "easy" with that system, at ultra/high settings. Maybe if you mean average, not minimum, it could do it.

 

[Zodiark1593]

I still run a 1045 at 3.5GHz and score 60FPS easy at 1080p in titles such as Witcher 3, Fallout 4, etc ... with ultra/high settings with a 960 4GB card.

I'm effectively gonna get to skip the entire Bulldozer series of processors until Zen thanks to Phenom II X6.

Kinda off topic, but judging by all the hate on the 960, it should be pulling 30 fps.

 

[Zor Prime]

Well since it sparked some nostalgic interest here's my sys specs:

1045 Thuban core PII X6 @ 3.45 GHz (I was wrong apparently) still a bump from the stock 2.7 is decent. It can and will run higher, up to around 3.8 semi-stable but I'll get BOINC computational errors in Rosetta, so, back down it goes. At 3.45GHz I can actually undervolt it from stock volts! It does not like being on high volts, it just doesn't really care.

8GB of that kickass Samsung 1600 DDR3 RAM that was being sold for while, @ 1700MHz 9-9-9-24. SINGLE CHANNEL, BTW, as my board decided one day two of the DIMM slots don't work anymore.

I do use a 256GB SSD for my games and swap disk, I keep a few loaded on it at any given time, among my 7TB worth of random SATA/USB3 drives.

Yes 60FPS isn't minimum but average, however it stays average probably about 99.x% of the time -- so.

I decided to bench Shadow of Mordor which is a very VRAM demanding game, I believe.

Note, I think synthetic benches while entertaining are shit. I'm a real person, so I did a Real World(tm) bench. I had Firefox loaded, a GRIDCOIN wallet going, 20% of my 8GB of my (single channel) RAM was preoccupied with random things. Oh, and my 960 is the EVGA FTW 4GB card which is the top 960 they sell. I chose it for efficiency on power -- I wanted a 970, but I can't pick a crippled, half-cobbled card I'd have to go for the 980.

I did two or three runs in three different GFX configurations:

Ultra, Bench 1
Av. 54
Max 87
Min 22

Ultra, Bench 2
Av. 54
Max 157
Min 39

Ultra (Shadow on High) Bench 1
Av. 64
Max 263
Min 32

Ultra (Shadow on High) Bench 2
Av. 64
Max 255
Min 46

Note, there were no spikes unless the frames shot through the roof. The min is within milliseconds of beginning the benchmark itself.

The differences between run 1's and 2's? Overall positive, but a third time exhibits cessation of any further noticeable benefit. It seems between run 1 and 2, on run 2, assets are pre-loaded in VRAM or some such thing ... ? Not sure, no clue.

I did a high GFX run across the board, still, better GFX likely than say, PS4.

High 1
Av. 67
Max 156
Min 49

High 2
Av. 68
Max 257
Min 51

High 3
Av. 67
Max 249
Min 51

I attribute the third run having lower numbers due to Real World(tm) processes taking their toll in the background.

Hope this is intriguing.

 

[TeknoBug]

I still have a Phenom II X6 1090T sitting here but no functional motherboard atm, I don't have my FX 4350 system here with me right now to throw that into.

And the i3 6300 experiencing stuttering? that's strange and I don't remember the last time anyone complained about stuttering on an i5 2500K which is still a strong processor today.