Thinking of switching from Intel to AMD

alex_123_fra · 16 Nov 2013 at 14:07

almighty15 said:
No point in running anything in the Cinebench R15 thread as Intel are pretty much ruling every result, Don't have BF4 as I'm not a BF fan...

If anyone has an FX then drop me message we'll decide and run some benchmarks..

Who mentioned cinebench? We are talking games here man. On Cinebench R15, we have already shown that overclocked 8350s and 8320s reach the sorts of multithreaded performance levels similar to whyscotty's i7 2600k at 4.8 GHz.

So you don't have BF4, fine. Tell us what games you have (especially multithreaded ones). Then we can see who has what and run some comparisons.

Martini1991 · 16 Nov 2013 at 14:07

No that's not better

Maximus VI Hero + 4670K

Asus Crosshair V + FX8350

That's what I mean.

You can go cheaper on the AMD with the FX8320.

Jeez.

almighty15 · 16 Nov 2013 at 14:10

alex_123_fra said:
Who mentioned cinebench? We are talking games here man. On Cinebench R15, we have already shown that overclocked 8350s and 8320s reach the sorts of multithreaded performance levels similar to whyscotty's i7 2600k at 4.8 GHz.

So you don't have BF4, fine. Tell us what games you have (especially multithreaded ones). Then we can see who has what and run some comparisons.

Trying to cherry pick situations now are we? And....

Cinebench R15 Thread

4.8Ghz 2600k = 847 points
5Ghz FX 8320 = 803 points

2600k is 6% faster even though it's clocked 4% lower...

Martini1991 · 16 Nov 2013 at 14:13

Well, Multi-threaded is more than 1 thread, so that's not really cherry picking, so there's a plethora of games

I can't even think of many single threaded games.

nikki5974 · 16 Nov 2013 at 14:13

I got 841 on my FX on CB R15

teppic · 16 Nov 2013 at 14:14

Comparing clock speed between different architectures is rather pointless.

And now the 8320 is being put against an i7 when convenient, and i5 when convenient.

nikki5974 · 16 Nov 2013 at 14:16

843.Core i7 [email protected] 4cores/8 threads - Whyscotty
[email protected] 4modules/8 threads - nikki5974
[email protected] 4modules/8 threads - alex 123 fra
827.Core i7 [email protected] 4cores/8 threads - b 2theenneh

Got some scores for ya from last page as you missed them

teppic · 16 Nov 2013 at 14:17

Martini1991 said:
Well, Multi-threaded is more than 1 thread, so that's not really cherry picking, so there's a plethora of games

I can't even think of many single threaded games.

I could write one for you in C, a guess the number game

While it's true the majority of games don't use 8 cores, the majority do use more than 2. It's really only games designed to hammer two threads where there's any issue (and those are typically the games mentioned, like Star Craft II).

Martini1991 · 16 Nov 2013 at 14:18

Those 2 i7 results are confusing, the 3770K at 4.6GHZ should be about the same as the 4.8GHZ 2600k :confused:

alex_123_fra · 16 Nov 2013 at 14:19

almighty15 said:
Trying to cherry pick situations now are we? And....

Cinebench R15 Thread

4.8Ghz 2600k = 847 points
5Ghz FX 8320 = 803 points

2600k is 6% faster even though it's clocked 4% lower...

Yes as said, I think you missed the fact that the Cinebench thread has not been updated by CAT for a while. So as Nikki says, see above ^

So not cherry picking at all. I hear no mention of what games you have. Or does the word multithreaded make you anxious :confused:

almighty15 · 16 Nov 2013 at 14:21

nikki5974 said:
843.Core i7 [email protected] 4cores/8 threads - Whyscotty
[email protected] 4modules/8 threads - nikki5974

Got some scores for ya from last page as you missed them

500Mhz clock speed advantage, double the cores and it still can't beat an Intel quad core with HT.

And the best thing is 4.8Ghz a very common clock on a 2600k's, the FX 's clock on the other hand very rarely gets achieved

nikki5974 · 16 Nov 2013 at 14:22

Martini1991 said:
Those 2 i7 results are confusing, the 3770K at 4.6GHZ should be about the same as the 4.8GHZ 2600k

Ram speed and timmings play a part in cinebench r15

nikki5974 · 16 Nov 2013 at 14:23

almighty15 said:
500Mhz clock speed advantage, double the cores and it still can't beat an Intel quad core with HT.

i think if you look at it this way the 2600k is clocked at a higher % than the FX

3.4 (i believe stock is ) to 4.8 = 1400 increase

the fx 4.0 - 5.3 = 1300 increase

my 841 score was clocked @ 5.4 which is same % overclock as the 2600k and only 2 points behind

Martini1991 · 16 Nov 2013 at 14:25

nikki5974 said:
Ram speed and timmings play a part in cinebench r15

It does, but it's a fair difference.

nikki5974 · 16 Nov 2013 at 14:26

Martini1991 said:
It does, but it's a fair difference.

maybe the guy didnt change his xmp in bios and was running the ram @ 1333

Martini1991 · 16 Nov 2013 at 14:26

nikki5974 said:
i think if you look at it this way the 2600k is clocked at a higher % than the FX

3.4 (i believe stock is ) to 4.8 = 1400 increase

the fx 4.0 - 5.3 = 1300 increase

my 841 score was clocked @ 5.4 which is same % overclock as the 2600k and only 2 points behind

That's a skewed way of looking at it, with a far from common overclock.

But it's moot.

almighty15 · 16 Nov 2013 at 14:26

nikki5974 said:
i think if you look at it this way the 2600k is clocked at a higher % than the FX

3.4 (i believe stock is ) to 4.8 = 1400 increase

the fx 4.0 - 5.3 = 1300 increase

my 841 score was clocked @ 5.4 which is same % overclock as the 2600k and only 2 points behind

Makes no difference at all, Not Intels fault AMD have to clock higher to try and compete.

nikki5974 · 16 Nov 2013 at 14:29

anyway we should be comparing i5 not i7 this is more like it

[email protected] 4modules/8 threads - Orch
[email protected] - Pazza
[email protected] 4modules/8 threads - ouzodan
[email protected] 4modules/8 threads - GinG
[email protected] 4modules/8 threads - Pazza
735.Core i5 [email protected] 4cores/4 threads - pastymuncher
731.Core i5 [email protected] - Martini1991
[email protected] 4modules/8 threads - ouzodan
712.Core i7 [email protected] 4cores/8 threads - smilertoo
695.Core i5 3570k@5Ghz - RavenXXX2
665.Core i5 [email protected] - Make
621.Core i5 [email protected] - Frozennonva

Martini1991 · 16 Nov 2013 at 14:30

It's also two gens old.
It's a pretty moot discussion, and you're just using flawed arguments which frankly make 0 sense.

The fact that Sandy stuff has a higher overclock percentage is a positive to the Sandys, as benchmarks compare stock results, and you've got more in the tank of the Sandy.
It doesn't mean sweet F all really. Sandy and PD generally clock to the same clock speeds, the sandy results just gain more.

Martini1991 · 16 Nov 2013 at 14:35

If we're going to actually compare multi-threaded Cinebench results, then use some common sense.

The single - multi-threaded performance of an i5 4670K shows its best and worst case scenario, however it only needs 4 threads to get to its best case scenario.

Whereas the FX83? It's worse at its worst case scenario, and it requires 8 threads to be at its best.

What's the common thread count on CPU's? It isn't 8, and situations that you use 8 threads (While greater than what it used to be) is in the minority.
So in a 1 threaded app, the 2500K is at 25% of its power, the FX83 at 12.5%
In 2 threads, it's 50% to the 2500K, 25% to the FX83.
3 threaded, 75% 2500K, 37.5% to the FX8350

And so on and so forth.

The i5 offers consistent performance (But lower raw performance) over an FX83 which offers inconsistent performance (But higher raw performance)
And gaming currently resides with lightly threaded games (There's a few examples of heavily threaded games granted, which will grow).