• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

First FX-9590 benchmarks

Page 4 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Status
Not open for further replies.
Poor AMD...

On the bright side, Jaguar's probably going to sell great. Not sure why AMD is worrying about e-Peen for enthusiasts when we all know that AMD's woefully behind here.
 
800MHz and about a 17% IPC advantage, but I could be wrong.

What kind of gpu usage are you getting? 20%?

What do you think?

9232503148_7a8bc3bf5c_o.png
 
Last edited:
Makaveli said:
Do you really think the 800 mhz difference in clock speed is making a 20fps difference in an Open GL benchmark when he is running at default clocks and i'm at Ghz speeds on our gpus?

hmm i'm leaning more towards drivers since his screenshot is using very old drivers will have to do some more testing.
Click to expand...

Yeah it was probably a bug in the drivers, AMD has come a LONG LONG way forward in the past 17 months in that regard. With synthetics like Cinebench, you can certainly see oddities. Very plainly that old 7970 at stock is not as fast as a Ghz or Ghz+ 7970.

CPU threaded bottlenecks are a factor, but not enough to cover 20fps.

It's also possible that if you had extra stuff running in the PCIe lanes, it might have polluted the results as well. Even if there is bandwidth to spare, I've found weird behavior at times with Physx cards in spare slots.
 
Power usage is fairly irrelevant on the mains IMO. I only care if that extra power usage results in superior performance at the same price. It looks like an overclocked 3930k is still the best performance for the money compared to the new $1k AMD chips, if they do end up at 1K.
 
Headfoot said:
Power usage is fairly irrelevant on the mains IMO. I only care if that extra power usage results in superior performance at the same price. It looks like an overclocked 3930k is still the best performance for the money compared to the new $1k AMD chips, if they do end up at 1K.
Click to expand...

Well not only that, but just a 5Ghz 2600k/2700k or 4.6Ghz 3700k or 4.4Ghz 4770k will easily beat it in almost everything, sometimes by a large margin, for a fraction of the price. On an AIR cooler with a midrange PSU/Mobo to boot!

In that respect, SB-E is really meant for very very specific uses, as most things can't properly take advantage of 6C/12T to a large enough degree to make up the lower OC headroom.
 
Arkaign said:
Well not only that, but just a 5Ghz 2600k/2700k or 4.6Ghz 3700k or 4.4Ghz 4770k will easily beat it in almost everything, sometimes by a large margin
Click to expand...

You re blindly wording assumptions that are unfounded.

The SB2600 is already at 4.3ghz in this already posted pic.

How much frequency to reach the FX score.?..

x264_FDH_Bench5GHz26HT_29fpsCopy_zpsa7bed35b.png


Once the things are threaded , because we dont buy 8C CPUs
to do quicktime dont we , things get tough for previous gen cores.
 
BallaTheFeared said:
800MHz and about a 17% IPC advantage, but I could be wrong.

What kind of gpu usage are you getting? 20%?

What do you think?

9232503148_7a8bc3bf5c_o.png
Click to expand...

llmm.png



Arkaign said:
Yeah it was probably a bug in the drivers, AMD has come a LONG LONG way forward in the past 17 months in that regard. With synthetics like Cinebench, you can certainly see oddities. Very plainly that old 7970 at stock is not as fast as a Ghz or Ghz+ 7970.

CPU threaded bottlenecks are a factor, but not enough to cover 20fps.

It's also possible that if you had extra stuff running in the PCIe lanes, it might have polluted the results as well. Even if there is bandwidth to spare, I've found weird behavior at times with Physx cards in spare slots.
Click to expand...

Fairly certain its drivers.

I will try disabling the NV card and redo the bench to see if there is any difference.

Do we see reviews from all the main sites for this AMD chip in the next few days?
 
Almost =! Everything 🙂

2700K @ 5Ghz gets about 26.4-26.5fps on that bench, which is pretty impressive for a 4C/8T CPU.

Encoding is the absolute best-case scenario for the FX, particularly with an old bench from over two years ago.

In the vast majority of apps/games, a 5Ghz SB/4.6Ghz Ivy/4.4Ghz Haswell will exceed the 5Ghz FX. All on air cooling for a fraction of the cost.
 
To put this into perspective just how laughable this is, my (four year old?) OC'ed Xeon W3520 (Xeon version of an i7 920, but I'm sure most in here know that) was faster than this cpu in the Fritz Chess benchmark and matched it's 3D Mark Vantage CPU score. I won't even go into how badly my present 3930K thrashes this thing, and combined they didn't cost as much as this thing is rumored to retail for....embarrassing.
 
InfoFront said:
Tigerdirect has the 9590 for $829.99.
http://www.tigerdirect.com/applicati...OUF.QjR.t3zNFg

The far more interesting chip to me is the 9370 for $319.99. If it sits roughly close to the 4770k in performance, that would be a fair price.
http://www.tigerdirect.com/applicati...SZncgTzqgjen5g

Also interesting is the fact that tigerdirect lists both parts as having 16mb of L2 and L3 cache.
Click to expand...

So it really is that expensive .... 🙂

Overclocking results will be worth reading. See what the 9370 can reach VS the 800 dollar gem.

They are adding the caches together. From the 8350 specs.

  • 8MB L3 Cache
  • 4 x 2MB L2 Cache
 
notty22 said:
So it really is that expensive .... 🙂

Overclocking results will be worth reading. See what the 9370 can reach VS the 800 dollar gem.

They are adding the caches together. From the 8350 specs.

  • 8MB L3 Cache
  • 4 x 2MB L2 Cache
Click to expand...

At 320$ the lower frequency part is not expensive at all
given that it s two bins over a 8350.

Edit : i dont expect it to reach the expensive part frequencies,
theses are individualy and lenghtily binned chips.

If the hefty one can do 5.3 then the cheap part can
do 5....
 
Last edited:
Abwx said:
At 320$ the lower frequency part is not expensive at all
given that it s two bins over a 8350.
Click to expand...

True, quite honestly at 4.7Ghz the $320 part is pretty reasonable, and that is particularly true for those who don't care to OC. It should look pretty good compared to a 4770K outside of power/heat/cooling aspects.
 
YBS1 said:
To put this into perspective just how laughable this is, my (four year old?) OC'ed Xeon W3520 (Xeon version of an i7 920, but I'm sure most in here know that) was faster than this cpu in the Fritz Chess benchmark and matched it's 3D Mark Vantage CPU score. I won't even go into how badly my present 3930K thrashes this thing, and combined they didn't cost as much as this thing is rumored to retail for....embarrassing.
Click to expand...

Using Fritz Chess for AMD chips has already been proven laughable, just as much as SuperPi.

Intel Chips are optimized in Fritz Chess.
 
I'm amused to still see Cinebench scores as a good view of what a processor can do in a 3D rendering scenario, when in real life applications things tend to be a lot more balanced:

This is a Vray bench, one of the most popular un/biased 3rd party 3D renderers for 3dsmax out there, an stablished standard in the arch viz scene:

IMG0041493.png


This is Mental ray, another renderer for 3dsmax that is also quite popular:

IMG0041492.png


I would really like this site would finally replace Cinebench for a more suited 3D renderer that really represents what happens in a real-life scenario, so we could finally put the "cinebench as best representation of FP performance in a CPU" mantra to rest.

PD: Im really impresed with HSW performance increase in Vray, a renderer I normally use. Seems the ppl at Chronos group could really make use of all the incremental increases made in the haswell core. And it also seems that the better the software is coded for multi-threaded performance, the less gains we start to see from HT, as the 4770 only is giving you 24% better render times for roughly 50%+ price compared to the 4670.
 
Last edited:
InfoFront said:
Tigerdirect has the 9590 for $829.99.
http://www.tigerdirect.com/applicati...OUF.QjR.t3zNFg

The far more interesting chip to me is the 9370 for $319.99. If it sits roughly close to the 4770k in performance, that would be a fair price.
http://www.tigerdirect.com/applicati...SZncgTzqgjen5g

Also interesting is the fact that tigerdirect lists both parts as having 16mb of L2 and L3 cache.
Click to expand...

Thanks for links!

Guess AMD doesn't expect these chips to last too long?
Anyone notice the warr..

Warranty provided by: AMD
Limited Warranty: 12 months parts; 12 months labor

EDIT FWIW, some old x264 scores from XS.org
d9i.png
 
Last edited:
Status
Not open for further replies.

TRENDING THREADS

Back
Top