Re: How fast are different ARM CPUs?
 

I am impressed by the smartq7 numbers, knowing it's only a little more than $200.

The smartq v7 numbers would be interesting too :)

Re: How fast are different ARM CPUs?
 

Is there an explanation for that ? There is a 1.2 GHz Marvell Kirkwood SoC inside the SheevaPlug. Is the OMAP3 that much more efficient ? I am on the impression that you guys are not surprise by that (are you ?).

Have you been able to find a performance per watt comparison ?

In short : :confused:

Re: How fast are different ARM CPUs?
 

Short version: ARMv5 cores (such as in the tablets, the SheevaPlug and the SmartQ) are scalar - they execute one integer operation per clock.

ARMv7 cores are superscalar - they execute multiple operations per clock (in the case of Cortex A8 that is in the N900 and Snapdragon that is in the N1, maximum two per clock). Since interdependence between instruction does not allow actual execution of two operations every clock, the ratio is about as is expected.

Re: How fast are different ARM CPUs?
 

Added Atom330 and Q9550 (Intel CPUs) numbers.

TA-t3: I believe it's all integer. Seems like along time since anybody worked on the arm cores for distributed.net, the most modern one was X-scale (which would favour Sheeva).

pierro78: Well if you're going to crunch RC5 or OGR, sure. But in your hands the SmartQ7 "feels" 3 times slower than N800/N810.

Mandor: I am not that surprised. ARM's Cortex A8 is a big step up in performance. Though sheevaplug's kirkwood is supposedly dual-issue too. Maybe it's the branch prediction in A8 that helps it, who knows.

Matan: Tablet and SmartQ are ARMv6.

Besides, ARMv7 doesn't mean superscalar. The Cortex A8 ARMv7 implementation is, but snapdragon is a different implementation and will perform differently.

Re: How fast are different ARM CPUs?
 

v7 is basically 7 with improved video decoding hardware.

Re: How fast are different ARM CPUs?
 

It's a different soc though.. the video decode is in the cpu/soc... different ram speeds...

BTW, N900 and SmartQ7 have almost identical /hardware/ video decoding capabilities.. Of course, N900 should be vastly better at doing it in software than SmartQ7 :)

Re: How fast are different ARM CPUs?
 

I don't know if this is a valid test or not.. It would seem to be so, but here we go with mine: I have a Seagate Dock Star which is running Debian Squeeze and it only has 128mb of Ram, though in fairness it also has a gig of swap space on the hard drive. In addition, the OS is on the hard drive instead of on flash, sd or nand.

Drive is a 5400 RPM Seagate 500mb drive, nothing special.

Here we go:

root@debian:~# cat /proc/cpuinfo
Processor : Feroceon 88FR131 rev 1 (v5l)
BogoMIPS : 1192.75
Features : swp half thumb fastmult edsp
CPU implementer : 0x56
CPU architecture: 5TE
CPU variant : 0x2
CPU part : 0x131
CPU revision : 1

Hardware : Marvell SheevaPlug Reference Board
Revision : 0000
Serial : 0000000000000000
root@debian:~# bash -c "time pi 1048576 >/dev/null" 2>&1 | grep real
real 0m0.003s
root@debian:~# bash -c "time pi 1048576 >/dev/null" 2>&1 | grep real
real 0m0.003s

Here is my "unlocked" quad core x86 machine with 4gb memory:

bvogele@NZXT:~$ cat /proc/cpuinfo
processor : 0
vendor_id : AuthenticAMD
cpu family : 16
model : 5
model name : AMD Phenom(tm) II X4 B40 Processor
stepping : 2
cpu MHz : 800.000
cache size : 512 KB
physical id : 0
siblings : 4
core id : 0
cpu cores : 4
apicid : 0
initial apicid : 0
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 5
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm 3dnowext 3dnow constant_tsc nonstop_tsc extd_apicid pni monitor cx16 popcnt lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt
bogomips : 6027.48
clflush size : 64
cache_alignment : 64
address sizes : 48 bits physical, 48 bits virtual
power management: ts ttp tm stc 100mhzsteps hwpstate

processor : 1
vendor_id : AuthenticAMD
cpu family : 16
model : 5
model name : AMD Phenom(tm) II X4 B40 Processor
stepping : 2
cpu MHz : 800.000
cache size : 512 KB
physical id : 0
siblings : 4
core id : 1
cpu cores : 4
apicid : 1
initial apicid : 1
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 5
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm 3dnowext 3dnow constant_tsc nonstop_tsc extd_apicid pni monitor cx16 popcnt lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt
bogomips : 6027.85
clflush size : 64
cache_alignment : 64
address sizes : 48 bits physical, 48 bits virtual
power management: ts ttp tm stc 100mhzsteps hwpstate

processor : 2
vendor_id : AuthenticAMD
cpu family : 16
model : 5
model name : AMD Phenom(tm) II X4 B40 Processor
stepping : 2
cpu MHz : 800.000
cache size : 512 KB
physical id : 0
siblings : 4
core id : 2
cpu cores : 4
apicid : 2
initial apicid : 2
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 5
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm 3dnowext 3dnow constant_tsc nonstop_tsc extd_apicid pni monitor cx16 popcnt lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt
bogomips : 6027.84
clflush size : 64
cache_alignment : 64
address sizes : 48 bits physical, 48 bits virtual
power management: ts ttp tm stc 100mhzsteps hwpstate

processor : 3
vendor_id : AuthenticAMD
cpu family : 16
model : 5
model name : AMD Phenom(tm) II X4 B40 Processor
stepping : 2
cpu MHz : 800.000
cache size : 512 KB
physical id : 0
siblings : 4
core id : 3
cpu cores : 4
apicid : 3
initial apicid : 3
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 5
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm 3dnowext 3dnow constant_tsc nonstop_tsc extd_apicid pni monitor cx16 popcnt lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt
bogomips : 6027.87
clflush size : 64
cache_alignment : 64
address sizes : 48 bits physical, 48 bits virtual
power management: ts ttp tm stc 100mhzsteps hwpstate

bvogele@NZXT:~$ bash -c "time pi 1048576 >/dev/null" 2>&1 | grep real
real 0m0.001s

Re: How fast are different ARM CPUs?
 

redrum: um, I'm guessing you don't have that "pi" software installed. Try doing just "pi 1048576" without the quotes..

Re: How fast are different ARM CPUs?
 

i dont have access to one right now but if anyones got an Efikamx handy it would be a very good thing to see some speed stats for that
v7 Arm cortex A8/Neon 128bit SIMD Freescale i.MX51at 800MHz
799.53 BogoMIPS doesn't tell the whole story with ARM v7 it seems.

see the youtube video's

http://www.youtube.com/watch?v=qUQzl...eature=related
"OpenVG hardware acceleration by Freescale"

OpenVG overlayed on top of OpenGL 2 seems like a very nice flash scripting replacement If someone writes the code to use them in a generic video app using FFmpeg etc.

http://www.youtube.com/watch?v=pOSluW6vHL8
"Microsoft Word on ARM Powered Laptop using Genesi and Citrix solutions"

very cool for a cheap developer kit on the face of it, that ORA boot firmware sounds interesting as you dont need an expansive/expensive BSP

and it seems they are actually the one's getting and initially supporting the actual full Arm Linux porting of different distro's too as they did with the PPC linux before that.

http://www.powerdeveloper.org/forums...ic.php?p=13854

Re: How fast are different ARM CPUs?
 

there's also the latest Marvell ARM v7 A9 quad / NEON 128bit SIMD at 1.6 GHz with a shared 2 MB L2 cache memory on board to come to market Now
http://www.theregister.co.uk/2010/11...a_xp_arm_chip/

when someone gets their hands on a developer one that too would be a very nice thing to run real life tests on (the Arm Cortex compiled x264 AVC Encoder etc) to give us all a window into what's coming down the line in 2011.