Fudzilla: Bulldozer performance figures are in

Page 47 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Status
Not open for further replies.
BlueBlazer

BlueBlazer

Senior member
Nov 25, 2008
555
0
76
drizek said:
Honestly, AES-NI is one of the features I'm most looking forwards to with Bulldozer. I don't encrypt now because I don't want to sacrifice performance.
Click to expand...
According to TrueCrypt benchmarks found ( http://www.legionhardware.com/artic..._2500k_and_core_i7_2600k_sandy_bridge,10.html and http://www.techspot.com/review/353-intel-sandy-bridge-corei5-2500k-corei7-2600k/page11.html ), Core i5 2500K and i7 2600K scores are about the same. This and the benchmark figures contradicts Obrovsky charts. Also the score for BD contradicts the official slide http://h3.abload.de/img/3ncx5.jpg (indicated as ">1600", or "2X > 1100T").

RussianSensation said:
Ok but how many times are you going to do this task? Once you to encrypt your hard drive / partition and then the rest is easily managed in real time, right? 2500k gets 2.5+ GB/sec. So a 3TB hard drive will be finished in less than 30 minutes if my math is correct? Even if you have 30TBs, that's less than 5 hours of encryption. And after that? You are seriously telling me it's worth spending $300 for that 1 task over modern processors? Even if it took you 2 full nights (16 hours), that's still not a big deal, unless I am missing something?
Click to expand...
Take into account the speed read and write speed of current SATA interface, at 2.5GB+ encryption speed is an overkill (underutilized) unless using a PCI Express based SSD like http://www.engadget.com/2011/08/02/oczs-z-drive-r4-pcie-ssd-offers-2-800mb-sec-500-000-iops-pl/.
 
A

Abwx

Lifer
Apr 2, 2011
11,885
4,873
136
BlueBlazer said:
According to TrueCrypt benchmarks found ( http://www.legionhardware.com/artic..._2500k_and_core_i7_2600k_sandy_bridge,10.html and http://www.techspot.com/review/353-intel-sandy-bridge-corei5-2500k-corei7-2600k/page11.html ), Core i5 2500K and i7 2600K scores are about the same. This and the benchmark figures contradicts Obrovsky charts. Also the score for BD contradicts the official slide http://h3.abload.de/img/3ncx5.jpg (indicated as ">1600", or "2X > 1100T").
Click to expand...

4d50832c-df62-4899-8257-af5f7b7b1134.jpg




http://www.xbitlabs.com/articles/c [...] 90x_5.html
 
BlueBlazer

BlueBlazer

Senior member
Nov 25, 2008
555
0
76
Abwx said:
4d50832c-df62-4899-8257-af5f7b7b1134.jpg


http://www.xbitlabs.com/articles/c [...] 90x_5.html
Click to expand...
That's not the same benchmark quoted in Obrovsky's "benchmark" chart and AMD's official slide. There are two different benchmarks in TrueCrypt >> http://www.technic3d.com/review/cpu-s/1200-intel-core-i7-2600k-und-core-i5-2500k/7.htm and http://www.hardcoreware.net/intel-sandy-bridge-performance-review/9/ (one is "AES" and the other "Serpent-Twofish-AES") :p

From http://www.legionhardware.com/artic..._2500k_and_core_i7_2600k_sandy_bridge,10.html .....
Application_04.png


1100T scoring 770 which is similar to AMD's own slide showing 797. This one from http://www.hardcoreware.net/intel-sandy-bridge-performance-review/9/ is even closer, at 786MB/s. Even the scores in your charts (from xBit) are radically different. Anyone can easily deduce that the TrueCrypt benchmark is the first "AES" one. :)
 
Last edited:
A

Abwx

Lifer
Apr 2, 2011
11,885
4,873
136
BlueBlazer said:
That's not the same benchmark quoted in Obrovsky's "benchmark" chart and AMD's official slide. There are two different benchmarks in TrueCrypt >> http://www.technic3d.com/review/cpu-s/1200-intel-core-i7-2600k-und-core-i5-2500k/7.htm and http://www.hardcoreware.net/intel-sandy-bridge-performance-review/9/ (one is "AES" and the other "Serpent-Twofish-AES") :p

From http://www.legionhardware.com/artic..._2500k_and_core_i7_2600k_sandy_bridge,10.html .....
Application_04.png


1100T scoring 770 which is similar to AMD's own slide showing 797. This one from http://www.hardcoreware.net/intel-sandy-bridge-performance-review/9/ is even closer, at 786MB/s. Even the scores in your charts (from xBit) are radically different. Anyone can easily deduce that the TrueCrypt benchmark is the first "AES" one. :)
Click to expand...


The 2500 AND 2600 have same perf even the latter has 8 thread capability
and 3% higher frequency at least.

Quite surprising, since we see from the X4/X6 difference that
this bench see better specs with increased thread capability.
 
BlueBlazer

BlueBlazer

Senior member
Nov 25, 2008
555
0
76
Abwx said:
It doesnt contradict OBR s slide since AMD only said that it s more than 2X the perf of the 1100T without giving any absolute value number. :)
Click to expand...
SB scores are around 3.5X higher. If BD can do better than 2X, wouldn't AMD want to confidently highlight that (showing it can perform as fast or better than SB)? ;)
 
A

Abwx

Lifer
Apr 2, 2011
11,885
4,873
136
BlueBlazer said:
SB scores are around 3.5X higher. If BD can do better than 2X, wouldn't AMD want to confidently highlight that (showing it can perform as fast or better than SB)? ;)
Click to expand...

For the time , they make their best to ensure a complete black out
about BD perfs , so it wouldnt be surprising that they published
only a part of the picture in their slide..:)
 
BlueBlazer

BlueBlazer

Senior member
Nov 25, 2008
555
0
76
Abwx said:
For the time , they make their best to ensure a complete black out
about BD perfs , so it wouldnt be surprising that they published
only a part of the picture in their slide..:)
Click to expand...
This is as close as anyone can get to an official benchmark/performance indicator. However it only shows AES hardware acceleration and not a true indicator of performance. Logically speaking, by low-balling performance numbers (showing slower performance than competition) would simply drive more enthusiasts, upgraders and potential customers to go for Intel instead. Wouldn't AMD like to stop this "exodus"? ;)
 
A

Abwx

Lifer
Apr 2, 2011
11,885
4,873
136
BlueBlazer said:
This is as close as anyone can get to an official benchmark/performance indicator. However it only shows AES hardware acceleration and not a true indicator of performance. Logically speaking, by low-balling performance numbers (showing slower performance than competition) would simply drive more enthusiasts, upgraders and potential customers to go for Intel instead. Wouldn't AMD like to stop this "exodus"? ;)
Click to expand...

Showing good numbers while the cpu is not available would be an
unecessary move, safe for....Intel , who would have an idea about
the bar level setting.

On another note, AES encryption surely use only integer ops ,
so BD with its 16 parallele exe units should be largely on par with SB s
12 shared units unless the AES instructions are badly implemented.
 
R

Riek

Senior member
Dec 16, 2008
409
15
76
BlueBlazer said:
SB scores are around 3.5X higher. If BD can do better than 2X, wouldn't AMD want to confidently highlight that (showing it can perform as fast or better than SB)? ;)
Click to expand...

Just a thought:
Maybe they do by talking about the lowest FX part (FX4) and therefore are able to use the > sign. (because they know no BD will perform lower).
 
BlueBlazer

BlueBlazer

Senior member
Nov 25, 2008
555
0
76
Abwx said:
Showing good numbers while the cpu is not available would be an
unecessary move, safe for....Intel , who would have an idea about
the bar level setting.
Click to expand...
Showing good numbers (logically speaking) would put BD in a better light, unless the CPU is "non-existant", "not ready" or not performing as intended.

Abwx said:
On another note, AES encryption surely use only integer ops ,
so BD with its 16 parallele exe units should be largely on par with SB s
12 shared units unless the AES instructions are badly implemented.
Click to expand...
IMHO would have to say AES acceleration is either not as good or not as optimized as those found in SB. Or perhaps it scales with higher clocks (than the ones AMD had at that time), just a speculation. ;)

Riek said:
Just a thought:
Maybe they do by talking about the lowest FX part (FX4) and therefore are able to use the > sign. (because they know no BD will perform lower).
Click to expand...
It can be interpeted that way, or comparing it to an "equivalent" 6-core FX part (FX6 series) since it was comparing with 1100T. But then again, why would AMD want to try low-balling performance numbers that would send all the wrong signals (to potential customers)? Is there a sense of logic here? :p
 
I

Idontcare

Elite Member
Oct 10, 1999
21,110
64
91
Abwx said:
Showing good numbers while the cpu is not available would be an
unecessary move, safe for....Intel , who would have an idea about
the bar level setting.

On another note, AES encryption surely use only integer ops ,
so BD with its 16 parallele exe units should be largely on par with SB s
12 shared units unless the AES instructions are badly implemented.
Click to expand...

1. Assuming software compiled for AES has no Intel-only "optimizations"
2. Assuming there are no critically limiting deficiencies in the cache structure of BD (latency and bandwidth)
3. Assuming there are no critically limiting deficiencies in the latency of the ISA implementation for AES on BD itself

Regarding (1) - this should not be happening and I would think that AMD would be the first to know and the first to make a stink over it if it were happening.

Regarding (2) - for whatever reasons, Intel's cache hierarchy has historically produced superior results in terms of latency and bandwidth...this could prove to be THE differentiating factor

Regarding (3) - AMD has traditionally chosen the path of implementing the most recent ISA expansions in a manner which can be generously referred to as "conservative" (lower performing) and follows up with more aggressive implementations in subsequent releases...this could prove to be a differentiating factor in AES performance between the Bulldozer (Zambezi) versus Bulldozer+ (trinity) cores
 
D

drizek

Golden Member
Jul 7, 2005
1,410
0
71
RussianSensation said:
Ok but how many times are you going to do this task? Once you to encrypt your hard drive / partition and then the rest is easily managed in real time, right? 2500k gets 2.5+ GB/sec. So a 3TB hard drive will be finished in less than 30 minutes if my math is correct? Even if you have 30TBs, that's less than 5 hours of encryption. And after that? You are seriously telling me it's worth spending $300 for that 1 task over modern processors? Even if it took you 2 full nights (16 hours), that's still not a big deal, unless I am missing something?
Click to expand...

The real time stuff is the problem. Copying files from an ssd to a raid0 can saturate a modern CPU at over 500MB/s. Think of something like video encoding where you are reading and writing a lot of data, and the processing time for encryption can slow down the encoding process.
 
S

soccerballtux

Lifer
Dec 30, 2004
12,553
2
76
drizek said:
The real time stuff is the problem. Copying files from an ssd to a raid0 can saturate a modern CPU at over 500MB/s. Think of something like video encoding where you are reading and writing a lot of data, and the processing time for encryption can slow down the encoding process.
Click to expand...

so don't encrypt it

the encryption performance is not really in 99% of the buyers' minds, nor will it be.
 
videoclone

videoclone

Golden Member
Jun 5, 2003
1,465
0
0
its so close now the wait is almost over everyone!! 3 more weeks and we will all be reading reviews on this new super long awaited overdue AMD REAL TRUE ALL NEW ( not next generation ) Design from the ground up!
 
frostedflakes

frostedflakes

Diamond Member
Mar 1, 2005
7,925
1
81
in b4 delay. :p

I'm really anxious to see an in-depth architectural discussion and benchmarks, though. Trying to keep my expectations reasonable and will probably pick one up unless it's a complete flop (another Phenom or something like that). Not expecting it to blow Intel away, as long as the flagship 8-core model is a solid upgrade over my X4 955 I'll be happy.
 
StinkyPinky

StinkyPinky

Diamond Member
Jul 6, 2002
6,977
1,276
126
AMD have a dog and we all know it deep down in our bones. They'd be leaking benchmarks by now if they have a winner.
 
JimKiler

JimKiler

Diamond Member
Oct 10, 2002
3,561
206
106
videoclone said:
its so close now the wait is almost over everyone!! 3 more weeks and we will all be reading reviews on this new super long awaited overdue AMD REAL TRUE ALL NEW ( not next generation ) Design from the ground up!
Click to expand...

Who wants the BD release more, AMD fanboys or Intel fanboys sick of all the hype?
 
Status
Not open for further replies.

TRENDING THREADS

COMPANY
RESOURCES
FOLLOW
Top Bottom