2013 Athlon X4 640 vs. 2013 FX-4300...

waltchan

Senior member
Feb 27, 2015
846
8
81
Just purchased what perhaps to be the final-edition Athlon II X4 Propus. According to serial # (8th & 9th digit, E3), it was manufactured on May 2013. Turns out the last and final Athlon X4 model available to public was 640 model, not 645 or 650. Most were produced in 2010 or 2011.

Interestingly, Athlon X4 640 remained in production throughout 2013 in limited quantities along with FX-4300, FX-6300, and FX-8320 at the same time. FX-4100 and FX-6100 ended production in late-2012.

Anything great to say with Athlon X4 Propus compare to FX-4300, and would you take full 4-cores in Athlon X4 with missing L3 cache or 2-modules, 4-threads in FX-4300, if you were shopping for a processor in September 2013, let's say?

$_57.JPG
 
Last edited:

Abwx

Lifer
Apr 2, 2011
11,855
4,832
136
That seems... hard to believe.

That s what was measured by Hardware.fr under Fritzchess MT, wich is more heavy than Cinebench.

That s why they kept it to measure power numbers, and despite it s outmatched by their more recent chess games wich are used in the CPU scores.

The numbers are in the link i provided, CPU power is extracted by measuring the current that run through the dedicated 12V rail, that s the same method that THG adopted recently :

IMG0043829.png


The number include the losses in the VRMs wich are in the 10% range.
 

waltchan

Senior member
Feb 27, 2015
846
8
81
So, Athlon II X4 640 was marketed as a cheaper solution to FX-4300 in 2013, as well as FX-4100 leftover in inventory. Phenom II X4, like 945, haven't seen one produced in 2013 yet. Only 2012 max. I did look at all of Athlon 645 and 650 at eBay, but they're all 2011 max.
 

escrow4

Diamond Member
Feb 4, 2013
3,339
122
106
Just buy an i3. If you don't do anything that requires a quad buy an i3 and forget about it. An Athlon II offers nothing over it.
 

waltchan

Senior member
Feb 27, 2015
846
8
81
Just buy an i3. If you don't do anything that requires a quad buy an i3 and forget about it. An Athlon II offers nothing over it.
Athlon II X4 is a i5 technically with full 4 cores while FX is a i3 with 2 modules and 4 threads. Who's better?
 

Soulkeeper

Diamond Member
Nov 23, 2001
6,732
155
106
Propus would have been better had it gotten the same tweaks that went into llano
my llano beats FX-4xxx in most benchmarks. It's essentially ~8% faster than propus per clock to do that however.
compared to thuban this 8% boost was canceled out per core/clock by the lack of L3, propus didn't get that advantage
propus tempted me for a time as a budget option back before the "fusion" platforms showed up with fm1
It's all old/slow now however :/
 
Last edited:

MiddleOfTheRoad

Golden Member
Aug 6, 2014
1,123
5
0
Architecturally, the Propus core was one of AMD's better designs -- as many have stated it had 4 full cores. But the clock speeds are just too low for today's applications and having a locked multiplier just really limits its appeal.

Passmark scores a Athlon II x4 640 around 3357. That's a pretty low score for a desktop quad core. And that's really all you can squeeze out of it with its locked multiplier.

An AMD FX 4300 manages a respectable 4654 in passmark -- plus like every FX it is totally unlocked. My highest overclock with any CPU was with a FX 4300 -- it overclocked far better than my early FX-8320. I managed a stable 5.1 Ghz on my 4300 with water cooling. So in my opinion, an Athlon II X4 has no chance against the performance of an FX-4300. An unlocked Phenom II could come close, though.
 

SPBHM

Diamond Member
Sep 12, 2012
5,066
418
126
Athlon II X4 is a i5 technically with full 4 cores while FX is a i3 with 2 modules and 4 threads. Who's better?

probably the FX? much higher clock to begin with
a phone CPU can also be a true quad core, while a dual core Celeron haswell will destroy it.
 

Abwx

Lifer
Apr 2, 2011
11,855
4,832
136
At that time a 3220 @ 3.3GHz would knock it down.

There s a 3240 in the graphs above, not sure that a 3220 would do better.

i3s should be definitly discarded as well as i5s unless one do one task at the time, or has previously tested the configuration to see if it suit his specific multitasking needs...

Tests done by Computerbase.de show that both i3/i5 (including the 6600k) perfs collapse if you run two tasks at the same time, one FP and the other Integer to make good measure.

This doesnt happen with i7s whatever the generation.

http://www.computerbase.de/2015-08/...gramm-multitasking-test-cinebench-plus-winrar
 

waltchan

Senior member
Feb 27, 2015
846
8
81
Architecturally, the Propus core was one of AMD's better designs -- as many have stated it had 4 full cores. But the clock speeds are just too low for today's applications and having a locked multiplier just really limits its appeal.

Passmark scores a Athlon II x4 640 around 3357. That's a pretty low score for a desktop quad core. And that's really all you can squeeze out of it with its locked multiplier.

An AMD FX 4300 manages a respectable 4654 in passmark -- plus like every FX it is totally unlocked. My highest overclock with any CPU was with a FX 4300 -- it overclocked far better than my early FX-8320. I managed a stable 5.1 Ghz on my 4300 with water cooling. So in my opinion, an Athlon II X4 has no chance against the performance of an FX-4300. An unlocked Phenom II could come close, though.
Don't forget price/performance ratio here. All the Athlon II X4 Propus are depreciating in prices so fast they're starting to be better buys than FX. They're approaching $20 starting shipped next year, while FX-4100 (cheapest FX) will struggle higher at $40 price minimum.

There's good number of X3 Rana triple-cores out there you can enable the disabled forth-core with no problem, and I even purchased some Athlon II X2 220 Regor dual-cores with two additional cores disabled for less than $9 shipped each.
 
Last edited:

Yuriman

Diamond Member
Jun 25, 2004
5,530
141
106
There s a 3240 in the graphs above, not sure that a 3220 would do better.

i3s should be definitly discarded as well as i5s unless one do one task at the time, or has previously tested the configuration to see if it suit his specific multitasking needs...

Tests done by Computerbase.de show that both i3/i5 (including the 6600k) perfs collapse if you run two tasks at the same time, one FP and the other Integer to make good measure.

This doesnt happen with i7s whatever the generation.

http://www.computerbase.de/2015-08/...gramm-multitasking-test-cinebench-plus-winrar

Interesting graphs, but a relevant question to ask is, what is the compression speed in WinRAR on each of these while running Cinebench? I'm willing to bet the discrepancy in scores is from how scheduling is handled with varying thread counts.
 

Abwx

Lifer
Apr 2, 2011
11,855
4,832
136
Interesting graphs, but a relevant question to ask is, what is the compression speed in WinRAR on each of these while running Cinebench? I'm willing to bet the discrepancy in scores is from how scheduling is handled with varying thread counts.

You have the scores in single tasks separately and the result when performed simultaneously.

Using the i3 as exemple CB MT score is 350 and Winrar compression time is 4mn30s when performed separately, in simultaneous runs CB MT score is 109 and Winrar compression time is 5mn58s.

Compression time increase by 32.6% while CB rendering time increase by 220%.

Winrar use integer exe ressources while CB use FP exe ressources, so it s not a case where the ressources used by an apllication are unavailable for the other.

Conclusion is that the i3 multithreading is innefficient when more than one application is used, it is efficient when a single application use several threads, and this is also the case for i5s.
 
Last edited:

TheELF

Diamond Member
Dec 22, 2012
4,027
753
126
Winrar use integer exe ressources while CB use FP exe ressources, so it s not a case where the ressources used by an apllication are unavailable for the other.

Conclusion is that the i3 multithreading is innefficient when more than one application is used, it is efficient when a single application use several threads, and this is also the case for i5s.

Cine not only uses integer but uses all of the integer commands that are available and so does winrar.

Who in his right mind uses benchmarking software to multitask anyway?
If a program can use all available cores than of course it will loose a lot of speed if you run something else at the same time,no matter how many cores you have.

Things change with the witcher since the game is only able to use 4 cores to a high degree so if you have more then 4 threads you don't loose that much in winrar,since the game starts with high priority you don't loose much in game performance even on the i3 ,so who cares how long the archiving is going to take.
 

DrMrLordX

Lifer
Apr 27, 2000
22,749
12,751
136
The locked multiplier of a Propus chip is hardly a limitation on its clockspeed. I had my x4 635 running @ 3.75 GHz before I killed it in a lapping accident. Just crank up the old htt/base clock and off you go!
 

Abwx

Lifer
Apr 2, 2011
11,855
4,832
136
Cine not only uses integer but uses all of the integer commands that are available and so does winrar.

Cinebench is pure FP, 70% of the instructions used are SSE2.
Integer commands will be eventualy used for data manipulations.

Things change with the witcher since the game is only able to use 4 cores to a high degree so if you have more then 4 threads you don't loose that much in winrar,since the game starts with high priority you don't loose much in game performance even on the i3 ,so who cares how long the archiving is going to take.

Yet games are Integer code based like Winrar, the difference is due to games not managing to fully load CPUs, if these were two Integer applications like winrar the outcome would be similar to the first test.

Edit : i would have thought that the i5 wouldnt be affected, yet only i7s seems to be immune to this perfs degradation.
 
Last edited:

Lorne

Senior member
Feb 5, 2001
873
1
76
The 640 is rated a 95w cpu, Not saying it cannot run lower.
Propus and Regor are great CPUs for the time, Now they work great for home built closet servers, They create little to no heat That one could maybe go passive cooling with a Noctou.

At one time I had a HTPC with an AII 250 under volt, I got it down to about 27w and about 35w watching an HD movie and burning a BR same time.

I am about to use a AII 640 in a NIX closet server, Under volt, I am going to try passive cooling.
 

monkeydelmagico

Diamond Member
Nov 16, 2011
3,961
145
106
Anything great to say with Athlon X4 Propus compare to FX-4300, and would you take full 4-cores in Athlon X4 with missing L3 cache or 2-modules, 4-threads in FX-4300, if you were shopping for a processor in September 2013, let's say?

If it was around $25.- then that would be something great to say about an elderly Athlon. Decent chip for it's time.... still capable of most typical user functions and some moderate gaming. Will usually overclock to around 3.4-3.6 too. Mobo's are super cheap.

Back in 2011 when I had no money I picked up a Phenom II X4 830. Very similar except for L3 cache and slightly higher base clock. I proceeded to OC and game on it for a couple years until some funds and the upgrade itch hit.
 

Leyawiin

Diamond Member
Nov 11, 2008
3,204
52
91
A Yorkfied Q9650 from 2008 is better. Hell, a Yorkfield Q9650 is only slightly worse than an FX-4300. Overclock it and that difference pretty much evaporates.
 

monkeydelmagico

Diamond Member
Nov 16, 2011
3,961
145
106
A Yorkfied Q9650 from 2008 is better. Hell, a Yorkfield Q9650 is only slightly worse than an FX-4300. Overclock it and that difference pretty much evaporates.

Sure if you already have one laying around or can get it for much cheaper than ebay average sold listings. Priced out a good LGA 775 mobo that can support a core 2 quad lately? Suddenly AM3(+) starts to look a bit more attractive......