5820K Power

[hajsveen]

How many Titan X's could a Core i7 5820K run before bottlenecking is a problem?

[CarterDavison]

It would run out of PCI Lanes first

[Lotus]

Depends on what you're trying to do. If you're trying to drive games at 1080p 144hz, the CPU would be a bottleneck even with just a single 980ti in most games, and the second 980ti would lead to a bottleneck in every game I've ever heard of although you'd be getting more than 144 fps in the less demanding games.

 

If you're trying to drive 4k at 60hz, the extra multi-threaded performance will go unused and it's not really better than an i5-4690k.

 

Remember, both CPU and GPU loads scale with fps, but only GPU load scales with resolution. Currently no CPU exists that can drive demanding games at 144 fps. The single-threaded performance just isn't there.

[Aleksbgbg]

Depends on what you're trying to do. If you're trying to drive games at 1080p 144hz, the CPU would be a bottleneck even with just a single 980ti in most games, and the second 980ti would lead to a bottleneck in every game I've ever heard of although you'd be getting more than 144 fps in the less demanding games.

 

If you're trying to drive 4k at 60hz, the extra multi-threaded performance will go unused and it's not really better than an i5-4690k.

 

Remember, both CPU and GPU loads scale with fps, but only GPU load scales with resolution. Currently no CPU exists that can drive demanding games at 144 fps. The single-threaded performance just isn't there.

 

I heard somewhere that anything i3< cannot bottleneck a single GPU?

 

However let's imagine I never heard that, then ... Xeon? Double Xeons?

[Lotus]

I heard somewhere that anything i3< cannot bottleneck a single GPU?

 

However let's imagine I never heard that, then ... Xeon? Double Xeons?

Nothing you're describing can do it. As you go up in price, you're getting higher multi-threaded performance. More cores, hyper threading, so on. The G3258 has the same single-threaded performance as the 5960x. Single-threaded performance is the limiting factor here since games must be run in real-time and are thread limited. Even a 5960x will bottleneck a single 980ti in most games at 1080p 144hz.

[CarterDavison]

Nothing you're describing can do it. As you go up in price, you're getting higher multi-threaded performance. More cores, hyper threading, so on. The G3258 has the same single-threaded performance as the 5960x. Single-threaded performance is the limiting factor here since games must be run in real-time and are thread limited. Even a 5960x will bottleneck a single 980ti in most games at 1080p 144hz.

"bottleneck"

 

 

 

[Lotus]

@CarterDavison: You're really cherry picking. No one turns those "advanced graphics" settings on in GTAV and if you notice, you're not even close to hitting 144 fps consistently. He's hitting 45-55 fps in GTAV, which was the only CPU intensive game in there. So yeah, a 5960x can totally hit 50 fps in GTAV and BF4 runs well on FX8 CPUs (meaning it doesn't require large single-threaded performance per FPS and can take advantage of the multiple cores). Big news. Try hitting 144 fps in GTAV and you'll understand my point.

[CarterDavison]

@CarterDavison: You're really cherry picking. No one turns those "advanced graphics" settings on in GTAV and if you notice, you're not even close to hitting 144 fps consistently. He's hitting 45-55 fps in GTAV, which was the only CPU intensive game in there. So yeah, a 5960x can totally hit 50 fps in GTAV and BF4 runs well on FX8 CPUs (meaning it doesn't require large single-threaded performance per FPS and can take advantage of the multiple cores). Big news. Try hitting 144 fps in GTAV and you'll understand my point.

He's hitting 55fps because it's completely maxed out at 8x msaa lol

 

Considering on my overclocked 980 and my i5 I only get 20fps completely maxed out with 8x msaa if i'm lucky lol.

[hajsveen]

He's hitting 55fps because it's completely maxed out at 8x msaa lol

 

Considering on my overclocked 980 and my i5 I only get 20fps completely maxed out with 8x msaa if i'm lucky lol.

 

 

@CarterDavison: You're really cherry picking. No one turns those "advanced graphics" settings on in GTAV and if you notice, you're not even close to hitting 144 fps consistently. He's hitting 45-55 fps in GTAV, which was the only CPU intensive game in there. So yeah, a 5960x can totally hit 50 fps in GTAV and BF4 runs well on FX8 CPUs (meaning it doesn't require large single-threaded performance per FPS and can take advantage of the multiple cores). Big news. Try hitting 144 fps in GTAV and you'll understand my point.

 

 

"bottleneck"

 

 

 

 

 

Nothing you're describing can do it. As you go up in price, you're getting higher multi-threaded performance. More cores, hyper threading, so on. The G3258 has the same single-threaded performance as the 5960x. Single-threaded performance is the limiting factor here since games must be run in real-time and are thread limited. Even a 5960x will bottleneck a single 980ti in most games at 1080p 144hz.

Would it work with only one GTX TitanX tho? 

[TheRandomness]

Would it work with only one GTX TitanX tho? 

A Titan X is a 980Ti with 12GB of VRAM.

[hajsveen]

A Titan X is a 980Ti with 12GB of VRAM.

So its enough cpu power?

[TheRandomness]

So its enough cpu power?

More that enough.

[Lotus]

It's not. No CPU can hit 144 fps in GTAV.

 

It's enough "power" in that sense, but it's too spread out over too many cores. The single-threaded performance just isn't there.

[Masada02]

It's not. No CPU can hit 144 fps in GTAV.

 

Challenge accepted.

[Lotus]

Challenge accepted.

Are you planning on turning CPU related settings down? That might change things enough to do it in certain situations.

[CarterDavison]

A Titan X is a 980Ti with 12GB of VRAM.

Not exactly..

 

Performs very similar tho

[TheRandomness]

Not exactly..

 

Performs very similar tho

980Ti performs better when it's aftermarket.

[hajsveen]

I havent mentioned gaming. I was just wondering if a 5820K would bottleneck a TitanX alot, or not that much.

[Aniallation]

I havent mentioned gaming. I was just wondering if a 5820K would bottleneck a TitanX alot, or not that much.

Well, what bottleneck situation are you talking about then? A Core 2 Duo won't bottleneck two Titan X in SLI if you're only watching YouTube videos.

[Masada02]

Are you planning on turning CPU related settings down? That might change things enough to do it in certain situations.

Turned down everything except population density.

 

Results:

Frames Per Second (Higher is better) Min, Max, Avg

Pass 0, 34.561775, 197.334869, 176.751205

Pass 1, 119.855827, 254.765442, 187.635971

Pass 2, 93.387131, 332.831238, 187.192780

Pass 3, 121.257996, 269.269836, 187.568069

Pass 4, 2.189468, 384.241364, 178.867416

 

CPU was only at 4.6 for this as well.

[CarterDavison]

 

Turned down everything except population density.

 

Results:

Frames Per Second (Higher is better) Min, Max, Avg

Pass 0, 34.561775, 197.334869, 176.751205

Pass 1, 119.855827, 254.765442, 187.635971

Pass 2, 93.387131, 332.831238, 187.192780

Pass 3, 121.257996, 269.269836, 187.568069

Pass 4, 2.189468, 384.241364, 178.867416

 

CPU was only at 4.6 for this as well.

 

roasted

[Lotus]

 

Turned down everything except population density.

 

Results:

Frames Per Second (Higher is better) Min, Max, Avg

Pass 0, 34.561775, 197.334869, 176.751205

Pass 1, 119.855827, 254.765442, 187.635971

Pass 2, 93.387131, 332.831238, 187.192780

Pass 3, 121.257996, 269.269836, 187.568069

Pass 4, 2.189468, 384.241364, 178.867416

 

CPU was only at 4.6 for this as well.

 

So view distance was at the minimum too right? I'm guessing that drastically changed this. Grats.

[Masada02]

So view distance was at the minimum too right? I'm guessing that drastically changed this. Grats.

I tested it with the distance scaling maxed as well and it didn't change much.

 

Frames Per Second (Higher is better) Min, Max, Avg

Pass 0, 23.787010, 248.189178, 156.406738

Pass 1, 87.064339, 283.366333, 185.868195

Pass 2, 104.963692, 243.942917, 174.869186

Pass 3, 6.772346, 249.587891, 185.727325

Pass 4, 49.632778, 268.379761, 174.405014

[Lotus]

 

I tested it with the distance scaling maxed as well and it didn't change much.

 

Frames Per Second (Higher is better) Min, Max, Avg

Pass 0, 23.787010, 248.189178, 156.406738

Pass 1, 87.064339, 283.366333, 185.868195

Pass 2, 104.963692, 243.942917, 174.869186

Pass 3, 6.772346, 249.587891, 185.727325

Pass 4, 49.632778, 268.379761, 174.405014

 

That's actually news to me. Thank you so much for this data. I'll look into this more, as it contradicts what people were claiming when the title launched.

 

edit: You're still hitting really low minimums. Averages are great though and above what I expected. Is this with the built-in benchmark tool or are you actually testing real in-game stuff?

[Masada02]

That's actually news to me. Thank you so much for this data. I'll look into this more, as it contradicts what people were claiming when the title launched.

 

edit: You're still hitting really low minimums. Averages are great though and above what I expected. Is this with the built-in benchmark tool or are you actually testing real in-game stuff?

That's just with the benchmark tool. Not sure where those minimums are coming from as they don't show up when the actual benchmark is happening. That may be in the first split second before each scene starts.