Monday, July 11th 2011

AMD FX-8130P Processor Benchmarks Surface

Here is a tasty scoop of benchmark results purported to be those of the AMD FX-8130P, the next high-end processor from the green team. The FX-8130P was paired with Gigabyte 990FXA-UD5 motherboard and 4 GB of dual-channel Kingston HyperX DDR3-2000 MHz memory running at DDR3-1866 MHz. A GeForce GTX 580 handled the graphics department. The chip was clocked at 3.20 GHz (16 x 200 MHz). Testing began with benchmarks that aren't very multi-core intensive, such as Super Pi 1M, where the chip clocked in at 19.5 seconds; AIDA64 Cache and Memory benchmark, where L1 cache seems to be extremely fast, while L2, L3, and memory performance is a slight improvement over the last generation of Phenom II processors.
Moving on to multi-threaded tests, Fritz Chess yielded a speed-up of over 29.5X over the set standard, with 14,197 kilonodes per second. x264 benchmark encoded first pass at roughly 136 fps, with roughly 45 fps in the second pass. The system scored 3045 points in PCMark7, and P6265 in 3DMark11 (performance preset). The results show that this chip will be highly competitive with Intel's LGA1155 Sandy Bridge quad-core chips, but as usual, we ask you to take the data with a pinch of salt.
Source: DonanimHaber
Add your own comment

317 Comments on AMD FX-8130P Processor Benchmarks Surface

#26
Pestilence
CasecutterI’d say that a 990FX mobo’s and FX-8130P together will be a much more competitive price package for the gaming enthusiast particularly once OC'd. Given AMD has had a price to performance lead in gaming against Intel up till now, I see Bulldozer as really becoming the preeminent platform for gaming. Especially considering the long term socket compatibility of AM3+, them coupled with Southern Islands, AMD has positioned it's self nicely.
How do you figure? Preliminary pricing has the 8 core BD at 330 dollars and the 990FX boards are priced around the same as P67 boards.

As for overclocking. Sandy Bridge processors are 95W TDP's, BD 8 core is 140W. Which do you think is going to have an easier time overclocking?
seronxZambezi*

and for those saying Bulldozer has weak core to core performance

It's a B1 Engineer Sample

Engineer Samples are to test functionality

B1 Engineer Sample =/= "B2" Consumer/Reviewer Sample
In performance

The scores will indeed go up from now to then when it releases
By 1% to 2% ser. There will be no miracle 10% gains.
Posted on Reply
#27
Thefumigator
seronxZambezi*

and for those saying Bulldozer has weak core to core performance

It's a B1 Engineer Sample

Engineer Samples are to test functionality

B1 Engineer Sample =/= "B2" Consumer/Reviewer Sample
In performance

The scores will indeed go up from now to then when it releases
Yeah it looks really good for an ES.
I personally blame the core to core performance "weakness" to the fact that in the test all 8 cores were used to the max, while a core to core performance should be tested using a single threaded application.

Maybe there's a memory bottleneck as they might be using slower DDR3, or the E. sample is configured for memory compatibility mode or something that degrades performance like that.

For those speculating about price, the chip being 32nm will make the price competitive and still be redituable for AMD.

Things look pretty good if those figures are for real.
Posted on Reply
#28
Pestilence
ThefumigatorMaybe there's a memory bottleneck as they might be using slower DDR3
They tested using DDR3 1866 so there's no bottleneck. Amd has always had shitty memory bandwidth
Posted on Reply
#29
DeerSteak
AMD certainly has a price advantage in motherboard costs. A 990X motherboard has the same 8+8 config for Crossfire/SLI that the P67 does, and AMD boards have better features than P67 boards at the same price point. And if you're not doing Crossfire/SLI you can certainly step down to a 970 board like I did and get one packed to the gills with 6x SATA III and 4x USB3 ports for $120. As for the CPU, AMD's quads are "fast enough" for games, but in sheer gaming performance AMD kind of universally loses to Intel. A $130 965 BE + (your favorite dedicated video card) loses to an i3 2105 + (same video card) in games most of the time.

A "fast enough" AMD CPU + mobo with nice amenities is roughly $250.
A "fast enough" Intel CPU + mobo with nice amenities is closer to $300.

In total system cost (if you're pairing it up with a $100 case, $80 PSU, $200 GPU, and $60 for 8GB of RAM) $50 isn't all THAT much. It's actually in non-gaming applications that AMD gets a performance lead with highly-threaded apps. Audio/video production, for example.
Posted on Reply
#30
LAN_deRf_HA
CasecutterI’d say that a 990FX mobo’s and FX-8130P together will be a much more competitive price package for the gaming enthusiast particularly once OC'd. Given AMD has had a price to performance lead in gaming against Intel up till now, I see Bulldozer as really becoming the preeminent platform for gaming. Especially considering the long term socket compatibility of AM3+, them coupled with Southern Islands, AMD has positioned it's self nicely.
With the cpu prices most likely being the same and the boards being out now we can get a pretty good idea that that's not accurate. At the cheap end ($120) the boards are comparative. At the premium side ($200) the boards are also comparable. The only part where they differ is the flagships are cheaper on AM3+. And I have no idea what you mean by "particularly once OC'd". The best bulldozer can do is match SB in overclocking. It'd be a bit absurd to expect BD to overclock better than SB. Lastly, AMD has only had a price/performance lead in gaming with Intel in the sub $100 category, and only because Intel is still relying on 775 for that price segment. AMD gaming performance is rather atrocious across the board compared to modern Intel platforms. They have to be dirt cheap to even match Intel in price/performance.
Posted on Reply
#31
DeerSteak
I gotta tell you guys, if you think per-core/per-clock performance for commercial CPUs is going to be significantly faster than that of engineering samples, you don't understand what an ES is for. The whole point of an ES is to ensure that the architecture is performing as expected.

OTOH, I'm very hopeful that default clocks are higher than 3.2GHz.
Posted on Reply
#32
Crap Daddy
The aida64 cache and memory benchmark are a disaster compared to Sandy Bridge, but again that might change,
Posted on Reply
#33
fullinfusion
Vanguard Beta Tester
Another fake posting. I thought all this crap was over with. I'm starting to just scroll past these useless threads. I just wish amd will release something solid for numbers already.
Posted on Reply
#34
Pestilence
DeerSteakI gotta tell you guys, if you think per-core/per-clock performance for commercial CPUs is going to be significantly faster than that of engineering samples, you don't understand what an ES is for. The whole point of an ES is to ensure that the architecture is performing as expected.

OTOH, I'm very hopeful that default clocks are higher than 3.2GHz.
I could have sworn the top of the line 8 core was going to be clocked at 3.8ghz
Posted on Reply
#35
seronx
ThefumigatorYeah it looks really good for an ES.
I personally blame the core to core performance "weakness" to the fact that in the test all 8 cores were used to the max, while a core to core performance should be tested using a single threaded application.
Core to Core performance is slightly weaker because of the new way the each core talks to each other in newer applications you'll see higher core to core performance
ThefumigatorMaybe there's a memory bottleneck as they might be using slower DDR3, or the E. sample is configured for memory compatibility mode or something that degrades performance like that.
L2 and L3 are the bottlenecks and those are getting fixed by B2
ThefumigatorFor those speculating about price, the chip being 32nm will make the price competitive and still be redituable for AMD.
$320~ish
ThefumigatorThings look pretty good if those figures are for real.
They are for real
PestilenceBy 1% to 2% ser. There will be no miracle 10% gains.
10 to 30% actually
PestilenceThey tested using DDR3 1866 so there's no bottleneck. Amd has always had shitty memory bandwidth
False, AMD has the fastest memory bandwidth
fullinfusionAnother fake posting. I thought all this crap was over with. I'm starting to just scroll past these useless threads. I just wish amd will release something solid for numbers already.
Not fake, just unofficial
Crap DaddyThe aida64 cache and memory benchmark are a disaster compared to Sandy Bridge, but again that might change,
Memory was fine they fixed the IMC but they might tweak it

L2 and L3 were below-par what they were expecting
PestilenceI could have sworn the top of the line 8 core was going to be clocked at 3.8ghz
FX-8130P 3.8GHz
FX-8110 3.6GHz
Posted on Reply
#36
DeerSteak
See, that's what I thought, too. 3.8GHz plus turbo is what I had *thought* i'd seen published in rumor/articles as well. That'd be totally bitchin'. Using the bizarre math that Apple used with the G5 quad was released (4x2.5GHz = 10GHz monster!!!11), we're looking at 30.4GHz of "CPU power". That's a retarded metric.

Regardless, 3.8GHz Bulldozer would be good news. Hell, I'd still OC the crap out of a 3.2GHz Zambezi and be happy.
seronxL2 and L3 are the bottlenecks and those are getting fixed by B2
Link? Also, the cache bandwidth tests already look pretty reasonable.
seronxFalse AMD has the fastest memory bandwidth
That's just...not true...based on every published review. For example: here
Posted on Reply
#37
Thefumigator
PestilenceThey tested using DDR3 1866 so there's no bottleneck. Amd has always had shitty memory bandwidth
Maybe 8 cores are too much for DDR3 1866. Even in a dual channel config.... again, maybe...
Posted on Reply
#38
repman244
Meh I still don't trust in these "leaks", need to wait for the real deal.
PestilenceAs for overclocking. Sandy Bridge processors are 95W TDP's, BD 8 core is 140W. Which do you think is going to have an easier time overclocking?
IF the leaked lineup from some time ago is correct FX-8130P 3.2GHz is a 125W chip, FX-8110 95W, I don't know where you got the 140W from.



And people that are comparing core for core, I don't think that's a valid comparison, since the "cores" in the modules aren't very similar to the real cores that we know (and AFAIK BD "core" has less transistors than Phenom II core).
Posted on Reply
#39
RejZoR
AMD's were never superior in SuperPI so why brag about that?
Posted on Reply
#40
DeerSteak
repman244And people that are comparing core for core, I don't think that's a valid comparison, since the "cores" in the modules aren't very similar to the real cores that we know (and AFAIK BD "core" has less transistors than Phenom II core).
While this is true, they need to be prepared for such comparisons since they've been "advertising" it (so to speak, when they discuss architecture) as 8 cores.

edit: they need to be honest and talk about it to the public as a 4-core with an enhanced form of SMT - there's more hardware duplication than HyperThreading, but it's not 8 full-blown CPU cores.
Posted on Reply
#41
seronx
DeerSteakLink? Also, the cache bandwidth tests already look pretty reasonable.



That's just...not true...based on every published review. For example: here
No, link you got to a trust a dude who has read up on Zambezi/Tech Specs for 7 months

Memory Bandwidth is higher and Latency is lower on AMD Chips always over Intel

AMD doesn't have a prefetcher/predictor

Zambezi does but how it works I don't know
DeerSteakWhile this is true, they need to be prepared for such comparisons since they've been "advertising" it (so to speak, when they discuss architecture) as 8 cores.

edit: they need to be honest and talk about it to the public as a 4-core with an enhanced form of SMT - there's more hardware duplication than HyperThreading, but it's not 8 full-blown CPU cores.
It is a full blown 8 core it is just that while they were looking away Intel caught up to CMP performance heck in CMP performance intel destroyed Phenom IIs just look at the i5 2500K vs Phenom II 980 BE
Posted on Reply
#42
Thatguy
repman244Meh I still don't trust in these "leaks", need to wait for the real deal.




IF the leaked lineup from some time ago is correct FX-8130P 3.2GHz is a 125W chip, FX-8110 95W, I don't know where you got the 140W from.



And people that are comparing core for core, I don't think that's a valid comparison, since the "cores" in the modules aren't very similar to the real cores that we know (and AFAIK BD "core" has less transistors than Phenom II core).
likely from sharing fpu, which leaves more room for int work.
Posted on Reply
#43
Pestilence
Ser,

10 to 30 percent from just a new stepping? Not a chance Imo.
Posted on Reply
#44
DeerSteak
seronxNo, link you got to a trust a dude who has read up on Zambezi/Tech Specs for 7 months
Surely unless you work for AMD or a partner (and if you want to claim to do so, that's fine) you've been reading what I've been reading, like David Kanter's and Jon Stokes' excellent write-ups.

Don't get me wrong. I really want Bulldozer to be competitive. Really, truly, with all my heart. But I'm not going to delude myself into thinking there's a magical performance improvement coming without higher clocks. If, when the reviews FINALLY come, there's extra performance to be had, I'm totally a winner since this is the platform I'm betting on with my own money.
Posted on Reply
#45
seronx
PestilenceSer,

10 to 30 percent from just a new stepping? Not a chance Imo.
I meant Engineer Sample to Consumer/Reviewer Grade Samples lol

Stepping will only be 5% at least 10% at most but could be 30% if there is more tweaks, by C0 30% will be hit absolute

Engineer is 3.2GHz
Consumer Grade is 3.6-3.8GHz(8110-8130P)

All FX Chips are unlocked so who cares about 8130Ps

I feel technically bad about the competition good luck overclocking your Multiplier Locked LGA 2011 4-cores and 6-cores


30% improvement in performance is only 55-60% over Phenom II 1100T in real world don't get your socks overblown yeesh

and to get back on the L2 and L3

They want the 2MB L2 to have the same latencies at a 256KB Intel L2 but with 8x the capacity
They want the 8MB L3 to do the same as the L3 is divided up in to portions of the module to help module communication
Posted on Reply
#46
DeerSteak
seronxNo, link you got to a trust a dude who has read up on Zambezi/Tech Specs for 7 months

Memory Bandwidth is higher and Latency is lower on AMD Chips always over Intel

AMD doesn't have a prefetcher/predictor

Zambezi does but how it works I don't know



It is a full blown 8 core it is just that while they were looking away Intel caught up to CMP performance
If this is a duplicate it's because I've written it before, and can't find it.

Do you work for AMD? You may have access to information I don't in that case. Otherwise I've been reading the same stuff as you - plenty of great writeups by incredibly smart people and nothing that I can find anywhere says anything that you're saying. Lots of people are reading and writing, and I'm more apt to believe what's been repeated over and over.

1.) Bulldozer in general is not a full 8 cores
2.) AMD's memory theoretical bandwidth numbers are lower than Nehalem and Sandy Bridge. And they're not THAT far off of theoretical maxes; roughly 25% short or so.
3.) They're just theoretical numbers. Nothing actually uses memory bandwidth like that anyway.

Particularly, Kanter's awesome writeup (the RealWorldTech link) goes to great lengths to say specifically the opposite of what you're saying. There's tons of intelligently-shared resources.
Posted on Reply
#47
Pestilence
seronx
I feel technically bad about the competition good luck overclocking your Multiplier Locked LGA 2011 4-cores and 6-cores
BD isn't competition for 2011 plus everyone is going to be buying the unlocked processors. I'd really love to see BD beat the 2600K badly that intel has to man up and release a 6 core on 1155 for 499.99 but i know i'm just dreaming.

Do you hear me Intel? We just want a 6 core. We don't NEED HT
Posted on Reply
#48
seronx
DeerSteak1.) Bulldozer in general is not a full 8 cores
2.) AMD's memory theoretical bandwidth numbers are lower than Nehalem and Sandy Bridge. And they're not THAT far off of theoretical maxes; roughly 25% short or so.
3.) They're just theoretical numbers. Nothing actually uses memory bandwidth like that anyway.
1.) Zambezi in general is a full native 8 core processor
2.) AMD's memory theoretical bandwidth numbers are substantially higher than Nehalem and Sandy Bridge
3.) Exactly but Applications can use AMD's memory to the theoretical limit unlike the competition

There is a big improvement to the NB and IMC in Zambezi a HUGE Improvement we won't see it till consumer grade
(From Phenom II "Black Edition" -> Zambezi "FX Black Edition Vision FX" Processors

I'm not an employee or work for AMD I just been looking a lot longer

PestilenceBD isn't competition for 2011 plus everyone is going to be buying the unlocked processors. I'd really love to see BD beat the 2600K badly that intel has to man up and release a 6 core on 1155 for 499.99 but i know i'm just dreaming.

Do you hear me Intel? We just want a 6 core. We don't NEED HT
Zambezi is damn as hell competing lol

The only LGA 2011 processor that is going to be unlocked is the Extreme Edition one($600-$1000)
Posted on Reply
#49
Crap Daddy
There's the i7-970 6 core 32nm for 580$. It is eating this leaked BD for breakfast not that anybody would need such a chip for gaming.
Posted on Reply
#50
DeerSteak
seronx, you're going to have to start linking stuff, because you're just making unsubstantiated claims. I've done the courtesy of linking everything I've claimed. Did you click on any of those links? They're ALL quoting directly from AMD and they're ALL saying that Zambezi is not 8 full cores.

I think your claim in #2 is that Zambezi supports DDR3-1866, and technically it does. The benchmarks show time and time again that it's irrelevant because Sandy Bridge + Nehalem both actually wring more bandwidth out of DDR3-1333 and DDR3-1600.
Posted on Reply
Add your own comment
Apr 15th, 2024 23:54 EDT change timezone

New Forum Posts

Popular Reviews

Controversial News Posts