AMD FX 8350 Winning against i5/i7?

[absolutezero]

I was on the market for a mid range CPU. I was originally gonna get a i5, but then I saw this video. This is very interesting results

Youtube Link

None Video Link: http://teksyndicate.com/videos/amd-...s-3820-gaming-and-xsplit-streaming-benchmarks

Tl;dr FX 8350> i5 9/10 times



Could this be a gpu bottleneck, should I stick with the i5? I am actually considering getting a HD 7870.

This disaster has gone on long enough. There's not enough Vicodin in the world to dull the pain.
-ViRGE

 

[inf64]

It's been already posted in other topic .

 

[ShintaiDK]

I think you should stick to the i5. Else I got some quality sand in Sahara I want to sell you.

 

[Chiropteran]

It's been already posted in other topic .

Could you be any more vague? What topic?

 

[Father Torque]

Stick with the i5 if you want less performance.

 

[inf64]

Could you be any more vague? What topic?

The ONLY other topic that has a 8350 in its title . Here you go.

 

[Centauri]

I think you should stick to the i5. Else I got some quality sand in Sahara I want to sell you.

What is that supposed to mean exactly? Your refutal to the results of a suite of conducted benchmarks is to... rebuke the results, and then just walk away?

Stick with the i5 if you want less performance.

Bingo.

 

[ShintaiDK]

What is that supposed to mean exactly? Your refutal to the results of a suite of conducted benchmarks is to... rebuke the results, and then just walk away?

Should be obvious how the FX performs in games if you check a broad varity of benchmarks. Including here at Anandtech.

Examples:
http://www.xbitlabs.com/articles/cpu/display/amd-fx-8350_6.html#sect0
http://www.anandtech.com/show/6396/the-vishera-review-amd-fx8350-fx8320-fx6300-and-fx4300-tested/5
http://www.computerbase.de/artikel/prozessoren/2012/test-amd-fx-8350-vishera/6/

The video/benchmark in the OPs link is simply game streaming with x264 from here: http://www.xsplit.com/

I dont think thats what the OP had in mind.

 

[AnonymouseUser]

The video/benchmark in the OPs link is simply game streaming with x264 from here: http://www.xsplit.com/

No, they are gaming while simultaneously streaming the game live using XSplit.

 

[guskline]

I have a 8350 with a GTX670 and it is FAST!

 

[SPBHM]

always try to look for benchmarks from more sources, and if you do that, it's pretty clear that for gaming (without xsplit streaming, which I do believe 99% of the gamers are not doing) the i5 is faster,

if you look at some of the results, there is something wrong with their testing (just look athe crysis warhead "without xpslit"), and how hard is to make a graphic to show the results on that blog post?

 

[boxleitnerb]

I'd rather believe a review with a well-documented set of benchmarks from people who know what they are doing than some dude on the web running benchmarks on youtube.
When additionally encoding in realtime, more cores will of course be beneficial, but this has nothing to do with gaming.

 

[Rvenger]

Canned benchmarks are canned benchmarks. Just saying... Most of all reviewers test the same way and doesn't really depict a long term real world experience.

 

[boxleitnerb]

Correctly done benchmarks very well depicpt real world experience. I'm with SPBHM on this. This guy did something wrong or didn't give all the information.

To take his Metro 2033 results as an example:

8350:

• 1080 - 36.44
• 1440 - 20.44

3570k:

• 1080 - 21.200
• 1440 - 12.800

3770K:

• 1080 - 27.480
• 1440 - 12.960

Performance drops dramatically from 1080p to 1440p -> GPU bottleneck. And yet he claims performance differences between the CPUs? Doesn't make sense at all.

Or look at the Trine 2 benchmarks. Significant performance increase due to hyperthreading, yet the engine is not significantly multithreaded according to Juha Hiekkamäki, who works at Frozenbyte.

These benchmarks are nonsense, end of story.

 

[StrangerGuy]

Yeah, everything you see on Youtube is real. Are you born yesterday?

 

[Arkaign]

Correctly done benchmarks very well depicpt real world experience. I'm with SPBHM on this. This guy did something wrong or didn't give all the information.

To take his Metro 2033 results as an example:

8350:

• 1080 - 36.44
• 1440 - 20.44

3570k:

• 1080 - 21.200
• 1440 - 12.800

3770K:

• 1080 - 27.480
• 1440 - 12.960

Performance drops dramatically from 1080p to 1440p -> GPU bottleneck. And yet he claims performance differences between the CPUs? Doesn't make sense at all.

Or look at the Trine 2 benchmarks. Significant performance increase due to hyperthreading, yet the engine is not significantly multithreaded according to Juha Hiekkamäki, who works at Frozenbyte.

These benchmarks are nonsense, end of story.

This.

The 8350 is a decent cpu, but these results are bogus.

 

[King4x4]

That video is just

 

[grimpr]

Lowering the resolution puts more workload on the CPU, try that with x264 running concurrently, see your fps flop.

 

[AtenRa]

These benchmarks are nonsense, end of story.

This.

The 8350 is a decent cpu, but these results are bogus.

Have you tried them your self and came up with this conclusion ??? or did you just made it up because those benchmarks show something different ??

 

[boxleitnerb]

What a surprise that you come out defending the 8350...
Look at the numbers, do some research, use common sense and you will come to the same conclusion. One doesn't need to test things themselves to find errors that others made.

 

[jones377]

Guys the results from the youtube video are from gaming AND streaming with Xsplit at the same time! This means realtime encoding of the entire screen while the game is playing. Encoding uses alot of CPU and the bigger the resolution, the more the CPU is taxed. Only the left over cycles goto the game itself, massively lowering the FPS you're getting compared with when you're not streaming. The extra cores on the 8350 and HT on the 3770k comes very much in handy here.

BTW, to my knowledge, Xsplit does not yet support QuickSynch. If it ever did, I would expect the 3770k in particular to do ALOT better than it does right now. This would be a perfect scenario for QuickSynch because it was intended for realtime streaming too and not just transcoding videos.

 

[boxleitnerb]

Yes, as far as I know, Xsplit is very integer heavy, that is why the FX shines, apart from the number of threads.
This has nothing to do with gaming in general.

 

[AtenRa]

What a surprise that you come out defending the 8350...
Look at the numbers, do some research, use common sense and you will come to the same conclusion. One doesn't need to test things themselves to find errors that others made.

I made a question, i havent defended anything.

jones377 said it very nicely, try gaming and streaming live using Xsplit at the same time and see for your self what happens, then make any conclusions.
I havent tried it and i will not make any conclusions until i have data. it is simple as that

 

[jones377]

I agree, a change in the topic title would avoid the confusion.

 

[jones377]

I would love to see Anandtech perform a similar test because the methodology and documentation performed by this guy in the video are not impressive at all... The results do seem to be all over the place and I heard no mention on the quality of the output stream. With such low FPS in some cases though, it looks like he gave Xsplit priority as far as CPU cycles are concerned though but that is just a guess.