9590 or 4770K?

[ALXAndy]

http://pclab.pl/art55318-3.html

Look how crap a 8350 is at 4.0Ghz shocking.

http://pclab.pl/art55318-4.html

5.0Ghz AMD8350 shafted by an i5 at 3.4Mhz, as originally posted above. (sorry for gpu limited post above)

Their results are so accurate that they don't show what level, nor is it even clear what the results are. What I mean is there are two sets of numbers. However they've not made it clear what they even mean (so I'm assuming they mean min and avg.)

The problem I have is that even if they are min and avg counts they don't add up at all with the numbers I got running my 8320 @ 4.7ghz with my 7990.

Secondly your 'hot as the sun' claims are just complete jibberish. If an AMD octo core gets to 68c it throttles. So even if you wanted to you couldn't make it even 70% as hot as Intel's finest. Don't use heat and power consumption as your argument, as the Haswell line of CPUs really do get truly hot. Like, really, really hot. That kind of scuttles your argument because that leaves you without a CPU to criticise. People would not be taking apart their CPUs (from Ivy and on) if they didn't need to.

The power argument? *sigh*. I've wasted over an hour of my time compiling data directly from energy suppliers (such as Eon and Scottish Power) and shown that running an octo core AMD for a year, four hours a day at full load (which is terribly unrealistic) you would spend £15 more per year.

Remember - science is going to absolutely roger you on this one. If a Haswell runs at 90c it is automatically wasting power by the logic of science. That heat is impossible to generate with low wattage and amperage. Waste heat = wasted power. At 90 c you're pumping energy into heat, wasting money. Oddly there don't seem to be any accurate power use benchmarks for Haswell at full pelt.

And then of course the argument works both ways. Why is the 8350 at 5ghz right up behind the 3770k. There's almost a £100 disparity between the two.

I would take a chill pill, man. If you have any requests then pop by my 8320 Analysis thread and ask for benchmarks.

 

[RNZZ]

deffo the i7

 

[gpuerrilla]

Well if you just attempted to quote the pro's and cons of both cpu's and not just their damn prices then people would not have to represent the opposite to your statements.

The 8350/9370 at 5.0Ghz is pulling way above 125w its at least 220w probably closer 300w when used in games like BF3/BF4 that max it out when played in crossfire/sli. (If your stupid enough to use an 8*** in crossfire/sli because it cripples your fps)

125w is probably is base setting at 4.0Ghz before it turbo's and certainly bears no relation to the 5.0Ghz overclocks.

Firstly it's relevant because a) Yes it is cheaper and b) it beats the intels when you stream and play as below quote will summarise the point of this thread. Clearly you need to take off those blinkers, bin them and read into Power.

There are plenty of people who are stupid according to you. Hmmm.

The power argument? *sigh*. I've wasted over an hour of my time compiling data directly from energy suppliers (such as Eon and Scottish Power) and shown that running an octo core AMD for a year, four hours a day at full load (which is terribly unrealistic) you would spend £15 more per year.

And then of course the argument works both ways. Why is the 8350 at 5ghz right up behind the 3770k. There's almost a £100 disparity between the two.

I would take a chill pill, man. If you have any requests then pop by my 8320 Analysis thread and ask for benchmarks.

Points above highlighted by me already in this thread - he cannot absorb text when he doesn't want to accept it. Deffo needs to take a chill pill, do they make 'fact' pills?

I was running a flipping Shuttle SFF system with a 975X motherboard,overclocked high VID Q6600,an overclocked 8800GTS 512MB(than a massively overclocked HD5850),4GB GB of DDR2 and multiple HDDs. That system lasted nearly 5 years. None of the parts were that energy efficient,so people crying over components in massive full sized ATX rigs is funny.

Forget that!!

Someone running GTX470 or GTX480 cards is worrying about power consumption!!

BAHAHAHAHAHAHA!!

Precisely - maybe that made him paranoid about consumption after he got his first quarterly (or his dad gave him a hiding for it).

It'd be interesting to see how that looks when streaming though, isn't that what the op wants?

See first quote above.

 

[teppic]

Secondly your 'hot as the sun' claims are just complete jibberish. If an AMD octo core gets to 68c it throttles. So even if you wanted to you couldn't make it even 70% as hot as Intel's finest. Don't use heat and power consumption as your argument, as the Haswell line of CPUs really do get truly hot. Like, really, really hot. That kind of scuttles your argument because that leaves you without a CPU to criticise. People would not be taking apart their CPUs (from Ivy and on) if they didn't need to.

Remember - science is going to absolutely roger you on this one. If a Haswell runs at 90c it is automatically wasting power by the logic of science. That heat is impossible to generate with low wattage and amperage. Waste heat = wasted power. At 90 c you're pumping energy into heat, wasting money. Oddly there don't seem to be any accurate power use benchmarks for Haswell at full pelt.

The heat is produced only via power consumption - a high core temp doesn't equal high power consumption though. It doesn't take a lot of energy to get a core temp of 90C (think of the temperature of the filament of a light bulb, over 2000C).

Core temps are pretty irrelevant ultimately as it doesn't relate to heat produced or energy used.

 

[Gimpymoo]

It'd be interesting to see how that looks when streaming though, isn't that what the op wants?

It is.

I am interested in what CPU would offer the best performance when streaming.

 

[gpuerrilla]

It is.

I am interested in what CPU would offer the best performance when streaming.

Then it's an i7 if you can afford it. Else FX8320 if your into value.

 

[dmferrett]

And this one goes against that completely:

http://gamegpu.ru/images/remote/htt..._GPU-Action-Battlefield_4-test-bf4_proz_2.jpg

The '5GHz' is actually 4.7GHz, they've made an error.

That data supports Pclabs findings and heres why, when reading off the table a 5.0Ghz 8350 is about parity with an i7 at 3.5Ghz lol. Thankyou berry much. (infact the 5.0Gghz amd is still slower)

 

[dmferrett]

The pclab results contradict pretty much everything else. There's no way that an 83xx CPU using all 8 cores well at 5GHz is going to do badly against an i5 at stock.

Alex's testing shows no real CPU bottleneck in BF4 with the 8320 OC'd, even with a 7990 (is that better, Martini? )

Even in Rome2 the 5.0Ghz AMD are performing terribly, seriously its a pity because I truely like AMD as a company and would follow them to the end's of the earth if they performed anything like an i5/i7 in games and clock for clock. (This part being important because gamers always try to max out and overclock their systems to get better fps, so if the ipc are poor then overclocking has very little net effect for a big thermal penalty)

 

[Lostmyrag]

It is.

I am interested in what CPU would offer the best performance when streaming.

4770k if you got the cash. It will be perfect, some of the best twitch streamers use this set up when using a single PC to stream. Otherwise you are looking at a 2 PC set up to achieve ZERO performance hit.

End.

 

[dmferrett]

Secondly your 'hot as the sun' claims are just complete jibberish. If an AMD octo core gets to 68c it throttles. So even if you wanted to you couldn't make it even 70% as hot as Intel's finest. Don't use heat and power consumption as your argument, as the Haswell line of CPUs really do get truly hot. Like, really, really hot. That kind of scuttles your argument because that leaves you without a CPU to criticise. People would not be taking apart their CPUs (from Ivy and on) if they didn't need to.

The power argument? *sigh*. I've wasted over an hour of my time compiling data directly from energy suppliers (such as Eon and Scottish Power) and shown that running an octo core AMD for a year, four hours a day at full load (which is terribly unrealistic) you would spend £15 more per year.

Remember - science is going to absolutely roger you on this one. If a Haswell runs at 90c it is automatically wasting power by the logic of science. That heat is impossible to generate with low wattage and amperage. Waste heat = wasted power. At 90 c you're pumping energy into heat, wasting money. Oddly there don't seem to be any accurate power use benchmarks for Haswell at full pelt.

Lets have a look at this more closely.

Everytime the cpu clocks or cycles a bit of work is done. The electrical energy is used in the transisitors to do processing work, which is all well and good. The down side is that for every cycle of processing there is heat given off as a by product. No electrical system is 100% Efficient.

Now as we increase the voltage in overclocked cpu's the heat per clock increases so our cooling requirements increase also, all sounds good so far.

The difference between and ivy/haswell compared to 8350 is that the transistor density is far greater and the physical properties of the Trigate transistors themselves inherently dissipate less thermal energy to the cooler. This is compouinded by the use of a TIM which doesn't move the heat to the cooler.

If you delid an IVY or Haswell people report 20oC drops in temperatures when overlocked. The heat in an Ivy/Haswell is a result of trigate and TIM otherwise the chip would run as cool and much more efficiently than an AMD clock per clock, simply becuase more processing is done per clock aka IPC is better. (This must make some sense even to the non scientist))

There is a difference between heat build up in the chip due to poor dissipation, and heat build up due to overclocking because the cpu is inefficient per clock.

The 32nm AMD transistors are less densely packed and dont use trigates yet and the TIM is better at transferring heat to the cooling system.

The heat is produced only via power consumption - a high core temp doesn't equal high power consumption though. It doesn't take a lot of energy to get a core temp of 90C (think of the temperature of the filament of a light bulb, over 2000C).

Core temps are pretty irrelevant ultimately as it doesn't relate to heat produced or energy used.

Core temps are important when you overclock due to thermal throttling.

 

[CAT-THE-FIFTH]

The FX8350 seems about Core i5 2500K/Core i7 2600K level(or slightly below it) in Rome II it seems:

http://gamegpu.ru/images/stories/Test_GPU/strategy/Total War ROME II/test/rome2 proz 2.jpg
http://gamegpu.ru/images/remote/htt...y-Total_War_ROME_II_Patch_2-rome2_p2_proz.jpg

Anyway,from someone who does actually play RTS games(instead of looking all day looking at charts),most of them after a few hours have terribad performance on systems whether they are Intel or AMD based(Intel might give 4FPS or AMD 2FPS),as they usually hit other issues,ie, like RAM limitations due to the nature of the programs. Its annoying at how many are not designed to handle 8+ hour multiplayer sessions in one go with six to eight players.

 

[dmferrett]

The FX8350 seems about Core i5 2500K/Core i7 2600K level in Rome II it seems:

http://gamegpu.ru/images/stories/Test_GPU/strategy/Total War ROME II/test/rome2 proz 2.jpg

Anyway,from someone who does actually play RTS games(instead of looking all day looking at charts),most of them after a few hours have terribad performance on systems whether they are Intel or AMD based(Intel might give 4FPS or AMD 2FPS),as they usually hit other issues,ie, like RAM limitations due to the nature of the programs. Its annoying at how many are not designed to handle 8+ hour multiplayer sessions in one go.

You mean my 200 hours in Rome 2 dont count!!!

But your right Rome 2 is such a poorly cpu bottlenecked game. I hope CA and Sega lose market share because the game is abismal. They have destroyed what would have potentially one of the best RTS games in history.

"SHAMEFUL DISPLAY"

 

[teppic]

That data supports Pclabs findings and heres why, when reading off the table a 5.0Ghz 8350 is about parity with an i7 at 3.5Ghz lol. Thankyou berry much. (infact the 5.0Gghz amd is still slower)

No, it doesn't support it in the slightest. Your bias is truly ludicrous.

 

[CAT-THE-FIFTH]

You mean my 200 hours in Rome 2 dont count!!!

But your right Rome 2 is such a poorly cpu bottlenecked game. I hope CA and Sega lose market share because the game is abismal. They have destroyed what would have potentially one of the best RTS games in history.

"SHAMEFUL DISPLAY"

Another game we play is Sins of a Solar Empire. Have had 8 mates around to play it and after we had literally hundreds(or thousands) of ships built after 12+ hours matches,it comes to a near halt. The Intel CPUs do a bit better than the AMD ones,but like with Supcom,the performance craters due to multiple reasons towards the end of long matches,RAM being one of them too. Its the same with so many RTS games(funnily enough SC2 is one of the better ones it seems).

The PC RTS games really need better testing by the companies who make them,as I get the impression they don't care,and at this rate they are going to kill the genre on the PC,and we will end up with crappy console versions,which will lack scope.

 

[dmferrett]

No, it doesn't support it in the slightest. Your bias is truly ludicrous.

Are you blind a 8350 at 5.0Ghz does not produce more frames than a 4770k at 3.5Ghz in BF4

Can someone else look at the graph please?

 

[teppic]

Are you blind a 8350 at 5.0Ghz does not produce more frames than a 4770k at 3.5Ghz in BF4

Can someone else look at the graph please?

In one instance you're saying that the 8350 at 5GHz is appalling next to a stock i5, now it's 'only' at parity with the 4770K, in two benchmarks that totally contradict each other.

 

[dmferrett]

In one instance you're saying that the 8350 at 5GHz is appalling next to a stock i5, now it's 'only' at parity with the 4770K, in two benchmarks that totally contradict each other.

The 4770k and 4750k are usually 1fps difference clock per clock.

A 5.0Ghz 8350 next to the i5/i7 at 3.5Ghz according to that benchmark in BF4,makes the AMD look woeful. Its there in black and white. Whats to argue about. I know first hand virtually all people with good performance that I play BF4 on line with 20 or so use Intel.

Like I keep saying if AMD was a vaible option I'd definately buy it over an Intel cpu.

All this textual bickering is pointless we both know which is best overall and which is cheapest overall.

 

[teppic]

The stock 8350 is performing equal to a stock i7 2600K there, and better than a stock 4670K. But those processors are fine, and the 8350 is bad. At 4.3 GHz (incorrectly listed as 4.7GHz) it's well ahead of both. And at 4.7GHz (listed as 5) it is within a couple of frames of the 4770K.

 

[dmferrett]

The stock 8350 is performing equal to a stock i7 2600K there, and better than a stock 4670K. But those processors are fine, and the 8350 is bad. At 4.3 GHz (incorrectly listed as 4.7GHz) it's well ahead of both. And at 4.7GHz (listed as 5) it is within a couple of frames of the 4770K.

Sorry which tables are you referring to?

 

[teppic]

http://gamegpu.ru/images/remote/htt..._GPU-Action-Battlefield_4-test-bf4_proz_2.jpg