AMD unleashes first ever commercial “5GHz” CPU, the FX-9590

[BallaTheFeared]

just don't mention the 3930K, $200 X79 boards and Haswell (and power usage)!!

lol their slides are always good for a laugh.

I know my i5 will already beat it's Cinebench score while using 1/3 the power :thumbsup:

 

[Vesku]

The only way this makes sense to me is if this is the end of the non-APU desktop CPUs for AMD. Otherwise they should have taken the money spent on this project and used it for pretty much anything else.

 

[galego]

It makes sense to me to fill the gap between Piledriver and Steamroller.

 

[RampantAndroid]

lol their slides are always good for a laugh.

I know my i5 will already beat it's Cinebench score while using 1/3 the power :thumbsup:

Also, there are numerous Z77 boards under $200. But hey, let AMD live with their head in the clouds for now....

 

[Remobz]

lol their slides are always good for a laugh.

I know my i5 will already beat it's Cinebench score while using 1/3 the power :thumbsup:

Poor AMD...sad indeed

 

[Vesku]

It makes sense to me to fill the gap between Piledriver and Steamroller.

The home heating gap?

 

[monstercameron]

The home heating gap?

because sli titans runs so much cooler...

 

[BallaTheFeared]

because sli titans runs so much cooler...

SLI Titans actually do something with that power though :thumbsup:

 

[Vesku]

because sli titans runs so much cooler...

Not thinking of SLI Titans thinking of moderately OCed (4.2GHz) 3770K. Here is my consumer math:

$180 FX 8350
+$50 Factory guaranteed OC
-$80 TDP increase of 100W

$150

Somehow I doubt that will be the MSRP.

 

[Chiropteran]

Not thinking of SLI Titans thinking of moderately OCed (4.2GHz) 3770K. Here is my consumer math:

$180 FX 8350
+$50 Factory guaranteed OC
-$80 TDP increase of 100W

$150

Somehow I doubt that will be the MSRP.

100W extra costs you $80, seriously?

What do you pay for power, like $5 per kwh?

 

[Vesku]

100W * 8760 hours (24 hours a day 365 day) = 876000 watt hours

876 kWh * $0.10 per kWh = $87.60 for 1 year continuous operation at the TDP difference (so assumes you use the exact workload that would closely approximate how AMD determines TDP)

 

[Hypertag]

As I have stated elsewhere Haswell=Hasbeen. It will barely compete against the coming Steamroller FX release. I know the proof is in the pudding, but the known slated improvements make steamroller a major upgrade over Vishera and close most of the gap in single threaded performance. In multithreaded I expect Steamroller to best I7-4770k without any doubt.

You have an account from 2001, and you have made seven posts?

In any event, you are correct. Failwell is the devil. It only offers more performance than its predecessor. Clearly, the AMD method of offering less performance and more power usage is the actual way of the future.

I have just purchased a few million in put options against Intel since you are clearly proven that Failtel will shortly declare bankruptcy. HEIL AMD!

 

[Hypertag]

100W * 8760 hours (24 hours a day 365 day) = 876000 watt hours

876 kWh * $0.10 per kWh = $87.60 for 1 year continuous operation at the TDP difference (so assumes you use the exact workload that would closely approximate how AMD determines TDP)

Frankly, there are very few areas of the US that have $0.10 per kWh anymore. $0.15 is a more accurate estimate for most individuals.

 

[Arachnotronic]

You have an account from 2001, and you have made seven posts?

In any event, you are correct. Failwell is the devil. It only offers more performance than its predecessor which offers more performance than its predecessor. Clearly, the AMD method of offering less performance and more power usage is the actual way of the future.

I have just purchased a few million in put options against Intel since you are clearly proven that Failtel will shortly declare bankruptcy. HEIL AMD!

Hypertag,

Intel makes more in net income in a quarter than AMD is worth. There's a reason Intel makes $$ and AMD hasn't known the meaning of "profit" for quite some time.

 

[Vesku]

Frankly, there are very few areas of the US that have $0.10 per kWh anymore. $0.15 is a more accurate estimate for most individuals.

When delivering estimates likely to be attacked by upset people I like to use numbers that are so conservative attacking them makes those people look silly. Since these estimates were used to come up with a discount on MSRP for a product I conserved in the downward direction.

 

[Hypertag]

The home heating gap?

But didn't you get the memos? Firstly, this processor will run at 60 degrees Celsius on AMD's useless TjCase sensor while Failwell will run at 90 degrees Celsius on the Core 3 internal temperature sensor. That alone proves that the AMD processor is better because. Secondly, as demonstrated by the experts in this forum, this processor will only have an actual TDP of 125 watts or 95 watts or something. AMD is being biased by setting the TDP of their processor at 220 watts because any 8350 can run at 4.7GHz while being undervolted.

 

[monstercameron]

Not thinking of SLI Titans thinking of moderately OCed (4.2GHz) 3770K. Here is my consumer math:

$180 FX 8350
+$50 Factory guaranteed OC
-$80 TDP increase of 100W

$150

Somehow I doubt that will be the MSRP.

you might be right but has anyone done the calculations comparing the money spent [sale price & power] to the amount of time it takes to get something done over a period of time?

 

[jhu]

Depends on the workload, an easier way to see the comparison would be here:

http://anandtech.com/bench/Product/434?vs=698

Despite the slower clock speed, the 8320 still beats the 8150. The average of BD vs PD at the same clock is about 8-9 % but games show about 15+% improvement on average. Not bad for a quick fix solution.

Well, ok, but 8320 has a higher turbo than 8150. But it's still nowhere near Phenom IPC. Take povray for example. 8320 does 4574 pps and phenom iix6 1090t does 3759 pps. For 8320 that's 163 pps/core/GHz if assuming it's not turbo or 142 pps/core/GHz if it is using turbo (i don't know if it is or not). In contract the 1090t is 195 pps/core/GHz if not using turbo, or 174 pps/core/GHz if it is using turbo (again i don't know if it is or not for that test). Either way, Phenom II still has higher IPC than Piledriver.

 

[tviceman]

Wow AMD just happily out Fermi'd Fermi.

 

[monstercameron]

Well, ok, but 8320 has a higher turbo than 8150. But it's still nowhere near Phenom IPC. Take povray for example. 8320 does 4574 pps and phenom iix6 1090t does 3759 pps. For 8320 that's 163 pps/core/GHz if assuming it's not turbo or 142 pps/core/GHz if it is using turbo (i don't know if it is or not). In contract the 1090t is 195 pps/core/GHz if not using turbo, or 174 pps/core/GHz if it is using turbo (again i don't know if it is or not for that test). Either way, Phenom II still has higher IPC than Piledriver.

wasn't the bulldozer/piledriver module approach about saving space to cram more coars into the system? cant quite remember but the numbers were like 50% space saving for 80% the performance of the phenom cores?

 

[jhu]

wasn't the bulldozer/piledriver module approach about saving space to cram more coars into the system? cant quite remember but the numbers were like 50% space saving for 80% the performance of the phenom cores?

That's fine, but the claim was that Piledriver has about Phenom II IPC. Clearly it does not.

 

[Vesku]

Well, ok, but 8320 has a higher turbo than 8150. But it's still nowhere near Phenom IPC. Take povray for example. 8320 does 4574 pps and phenom iix6 1090t does 3759 pps. For 8320 that's 163 pps/core/GHz if assuming it's not turbo or 142 pps/core/GHz if it is using turbo (i don't know if it is or not). In contract the 1090t is 195 pps/core/GHz if not using turbo, or 174 pps/core/GHz if it is using turbo (again i don't know if it is or not for that test). Either way, Phenom II still has higher IPC than Piledriver.

163/0.8 (CMT multithread penalty) = 203.75
142/0.8 = 177.5

(203.75+177.5)/2 = 190.6 average of the two

I'd say Piledriver is pretty close to Phenom II in terms of IPC.

 

[Enigmoid]

163/0.8 (CMT multithread penalty) = 203.75
142/0.8 = 177.5

(203.75+177.5)/2 = 190.6 average of the two

I'd say Piledriver is pretty close to Phenom II in terms of IPC.

Close but still a little behind.

 

[ninaholic37]

I feel IB-E may put the whole "debate" (really people think it's a technical issue and not cost based?) to rest. Pretty sure with their premium line they'll use solder or if adhesive and TIM they will have another solution to gap distance (bracket or what not).

Wonder if Kabini uses TIM, no expectations from enthusiasts so it would be an easy cost reduction.

I just assumed IB-E would use the same "tried and true" adhesive/TIM as IB. Since the cost and performance for the enthusiast chip is quite different though, who knows. Will be interesting to see what the real answer will be... :awe:

 

[guskline]

This is the chart I developed tonight using both of the rigs listed in my sig below. I ran the 8350 stock, the 3770k stock, the 8350 @4.7 Ghz which will be the base clock of the FX 9590 without the turbo to 5 Ghz and my 3770k @ 4.4Ghz which is what I run it at. I run my 8350 at 4.6 Ghz so I bumped it up a bit to replicate the approximate base of the FX 9590. Remember that the scores will likely be a bit higher due to the turbo of 5Ghz. I probably could have replicated that but it was easier just running all 8 cores at 4.7Ghz. I realize these are only cpu scores and hardly "all encompassing". I don't own a 4770k so no figures would be available. Perhaps a poster who owns one could help out. I ran all of the newest Aida64 benchmarks (own a licensed version), CineBench 11.5 and Passmark 8 CPU tests (own licensed version). I'm open to all other suggestions for comparison tests. At least with these figures one can make a comparison of some cpu tests between the FX 8350 stock, the I7- 3770k stock and the projected base of the FX 9590 base of 4.7Ghz. I threw in my figures for my I7-3770k @ 4.4 Ghz to also compare. Hope this helps the discussion.