AMD Athlon 5150 Kabini - cat, Jaguar or more fun?

FlanK3r · 17 Apr 2014 at 09:08

Im tetsing for review now Athlon 5150. This is from my point the best price/performance ration for AM1. Basic clock is 1600 MHz (top model 5350 has 2050 MHz, this different could be visible in practice, but 5150 has still OK performance). The RAM IMC support is 1600 MHz DDR3. Not bad, but you will see later. And iGPU has the core clock 600 MHz. For older oldschool games its OK. As Quake III. Quake IV, Counter Strike, Battlefield 2 or modern Torchlight.

New benchrig for small machine

And new HyperX memory Fure

But, what happens, if the right guy has the right motherboard? This motherboard is Asus AM1I-A.
Do not hotlink images - stulid.

BIOS is simply perfect for some fun.

This is teraser. 1600 MHz CPU and DDR3? No way...I wanna more... 1984 MHz at CPU and DDR3 the same! And still stock voltage of CPU

BIOS is overclockable, no problem for this great motherboard.

And Cinebench at stable settings:

Later coming more OC fun

Treznor · 17 Apr 2014 at 09:25

I've just got a 5350 and got it setup for file / plex media server, will be interesting to see what you can do!

I've not had time yet to play around with it, but so far I'm more than happy with the price of the kit.

cheesyboy · 17 Apr 2014 at 09:28

Nice work. Look forward to seeing more

It's good to see a >30% overclock, though with the 2.05Ghz 5350 being barely more expensive, I guess it's not much free performance.

FlanK3r · 17 Apr 2014 at 09:29

yes, now Im at performance of 5350 chip...And much higher in memory clock and iGPU clock

. But again, this board is gold. MSI and ASRock or Gigabyte have not BCLK in OC menu...

cheesyboy · 17 Apr 2014 at 09:34

FlanK3r said:
yes, now Im at performance of 5350 chip...And much higher in memory clock and iGPU clock . But again, this board is gold. MSI and ASRock or Gigabyte have not BCLK in OC menu...

For such a cheap board, it's good to see such features.

FlanK3r · 17 Apr 2014 at 12:58

Now Im at 2270 MHz

...

Jumper118 · 17 Apr 2014 at 16:56

this is brilliant its faster than my [email protected] and that is a 65w apu!

humbug · 17 Apr 2014 at 18:03

Such a good little chip.

Phenom II @ 4Ghz (twice the clock rate)

17.5 sec @ 4Ghz vs the Athlon 31.6 sec @ 2Ghz

It has roughly 10% better IPC clock for clock than Phenom II, or 15% better IPC than Vishera, and those are chips with L3 and various extensions the Athlon does not have.

The APU versions have even less IPC.

A full fat FX version of this chip @ 4Ghz would utterly destroy the FX83##, probable by something like 25% core for core clock for clock, and use much less power.

Martini1991 · 17 Apr 2014 at 18:08

I wouldn't make IPC comparisons with SuperPi....

humbug · 17 Apr 2014 at 18:11

Martini1991 said:
I wouldn't make IPC comparisons with SuperPi....

Why not? everyone else does.

Martini1991 · 17 Apr 2014 at 18:11

humbug said:
Why not? everyone else does.

No they don't (In fact, people explicitly state not to, you know because its dismal appearance on AMD, unless you're running a tweak), it's not relevant, hasn't been for a long long time.
Run Cinebench and I'm pretty confident we'll see a difference.

humbug · 17 Apr 2014 at 18:20

Martini1991 said:
No they don't (In fact, people explicitly state not to, you know because its dismal appearance on AMD, unless you're running a tweak), it's not relevant, hasn't been for a long long time.
Run Cinebench and I'm pretty confident we'll see a difference.

I'm pretty sure Cinebench is Integer performance and apparently still made with Intel's Compiler.

SupperPi is FP performance.

Martini1991 · 17 Apr 2014 at 18:21

Cinebench is Floating Point.

SuperPi, regardless of what it is, is irrelevant due to the code, Or that's what's been bleated for the past few years (Whenever people have compared in the past).

humbug · 17 Apr 2014 at 18:25

Martini1991 said:
Cinebench is Floating Point.

SuperPi, regardless of what it is, is irrelevant due to the code, Or that's what's been bleated for the past few years (Whenever people have compared in the past).

Ok

I can't be bothered going into the BIOS to turn off 2 cores but i can down clock to 2GHz and run a single core bench in the Asus Desktop clocking app.

brb..

Martini1991 · 17 Apr 2014 at 18:26

Just divide your score by 6 and then times it by 4

humbug · 17 Apr 2014 at 18:44

54. x4 = 216

So about 20% faster, we need a 2Ghz Atlon 750 to pit it against. i think.

humbug · 17 Apr 2014 at 19:57

I ran it @ 2Ghz on SupperPI, its the same application so it is comparable.

This little Athlon is 10% faster than Phenom II in SupperPI, despite the fact that the Phenom II is a Full Fat Desktop chip, i also doubt a Phenom II x4 @ 2Ghz is running anywhere near as low as 25 Watts.

I don't know whats going on in Cinebench, it needs to be compared with a similar cut down low end chip.

Its an impressive little chip no less.

Martini1991 · 17 Apr 2014 at 22:32

Despite the fact that for years people have been saying "Don't use Superpi" and "Superpi is useless for comparing" you want to use it, because it allows you to come out with statements which you can validate using the flaky Superpi.

This isn't a bash at Superpi itself, as it's useful for comparisons within the families, but comparing different series to another series is a no-no, and really is useless.

Like a bunch of Haswell people going at it, changing their RAM etc, then yeah, it's not bad at all.
Bunch of Phenom II lads going at it, yeah, despite the appearance of results, they can compare to each other.

But series to series? No.

I'm not knocking the Athlon in the slightest, but I don't agree with bigging something more than it needs to be, which is what you're doing with the "More IPC's than the Phenomssssss"

In its target market (Although, the power usage over its competition is being overlooked massively, as it's a bigger negative than review/enthusiasts are making out to be, given the market) the IPC of the Athlon trumps anything Intel has by a country mile, except Intel has higher clocks to compensate. In this case, AMD have also clocked up, but it's came at a higher power point. Ends in ultimately better performance however, but those who may look for these chips ; NUC's, that power usage is very important due to the TDP.

humbug · 17 Apr 2014 at 23:43

Martini1991 said:
Despite the fact that for years people have been saying "Don't use Superpi" and "Superpi is useless for comparing" you want to use it, because it allows you to come out with statements which you can validate using the flaky Superpi.

This isn't a bash at Superpi itself, as it's useful for comparisons within the families, but comparing different series to another series is a no-no, and really is useless.

Like a bunch of Haswell people going at it, changing their RAM etc, then yeah, it's not bad at all.
Bunch of Phenom II lads going at it, yeah, despite the appearance of results, they can compare to each other.

But series to series? No.

I'm not knocking the Athlon in the slightest, but I don't agree with bigging something more than it needs to be, which is what you're doing with the "More IPC's than the Phenomssssss"

In its target market (Although, the power usage over its competition is being overlooked massively, as it's a bigger negative than review/enthusiasts are making out to be, given the market) the IPC of the Athlon trumps anything Intel has by a country mile, except Intel has higher clocks to compensate. In this case, AMD have also clocked up, but it's came at a higher power point. Ends in ultimately better performance however, but those who may look for these chips ; NUC's, that power usage is very important due to the TDP.

Your Talking about Intel's Bay-Trail Celeron?

Your right, the Athlon CPU uses about 40% more power while being 20% faster, i didn't want to compare it to Intel, i don't know why that's so important to some people to do that, but on that note, this CPU does far better when compared to Intel than anything AMD have done since the first round of Athlon CPUs, its so much better and makes AMDs current FX and A series chips look like a joke, its a significant step forward in the right direction, thats got to be a good thing.
Now call me over optimistic, call me a fanboy if you like despite the fact that i'm in complete agreement with you on AMD's current line of CPUs;-
I don't want to focus everything i say on how negative it compares to the all glorious Intel.
I want to bring some focus on the improvements AMD are making and give it credit where its due, I know AMD have more work to do. i'm not going to spend my life "singing all hail Intel" because AMD haven't beaten them!

To much gloom and glory supporting around here, we are hardware enthusiasts, not Intel disciples. Well, i'm not....

Martini1991 · 17 Apr 2014 at 23:47

Why wouldn't you compare it to the Intel Cele Baytrail? It's the competing chip :confused:

And there was nothing "Hail Intel" about my post, it was just fact ; AMD's IPC trumps Intel in the mobile area, and has done for a while. Intels always compensated with clocks.
You can't call people Intel disciples, say we're hardware enthusiasts, when you're closing the door on caring about how the product stands to competition. Those who don't, are in fact the unbiased, because we don't only care about one offering, we see the picture.

And then I compared the product to how it'd be used within its market ; NUC's, and that TDP isn't great use for an NUC, but the product would be really good at a cheaper alternative to the i3 Intel NUC's. And with Steam Streaming working with Hardware Decoding? Well, you could run one of these Kabini's and you've got 1080p gaming in your living room.

Or, at least in theory, because you'd probably struggle doing an NUC passive solution on this chip.
So workstation NUC boxes for offices seems more likely, with active fan solutions etc.