• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

[Donanimhaber] AMD Trinity A10-5800K performance leaked

C

Chinoman

Senior member
http://www.donanimhaber.com/islemci...on-A105800K-Trinity-islemcisi-test-edildi.htm

CPU-wise it only seems like a marginal improvement over Llano. GPU-wise, this thing made a leap similar in proportion to Intel's Ivy Bridge IGP over Sandy Bridge.

ErPIE.jpg


Roughly equivalent to a Radeon 5670. Tremendous for an IGP.

With all the rebadging that has gone on in the lower-mid GPU market lately, I feel like this chip finally makes "buying discrete" obsolete in the sub-$500 build range.
 
Last edited:
That is pretty impressive. The Llano GPUs seemed like a nice step in the right direction, but still requiring another GPU for 1080p play for newer engines. Trinity looks like a solid budget build CPU. I think AMD will have the budget realm locked down if there are budget motherboards as well. It also will allow higher end crossfirex pairing with budget video cards. If they can keep crossfirex scaling well with onboard + dedicated cheap cards, trinity could fit a lot of different spots for budget builds.
 
These Fusion chips are easily the best thing AMD has going right now. Great products and they're doing much more to bring good graphics and in turn gaming to the masses. Intel however will drive the market so hopefully they keep trying to play catchup to AMD in the IGP department.
 
If the cpu performance is any indication of the improvement 'Piledriver' brings to the table, then we might as well forget the upcoming desktop cpus all together.:thumbsdown:
 
Vesku said:
Need to see more than pi test to make a judgement.
Click to expand...

+ Yes indeed, SuperPI is pretty useless. But compared to Llano's Star core, Piledriver's x87 is worse (per Mhz). The more interesting comparison is vs Bulldozer, but anyhow x87 is hardly something that AMD would bother to optimize for these days.

More importantly for cheapskates like me (currently running a S775 Pentium E5300 overclocked) is the price of the CPU and mobo. Current Llano's are poor value retail when the equiv. AMD + Radeon is cheaper, something's not right.
 
If they could at least match Clarksfields/Arrandales ST performance that would be a great chip for laptops. Alas that's not going to happen 🙁
 
Last edited:
There needs to be a more thorough test than just 3dmark and pi. 3Dmark looks great on paper, even with extremely bandwidth starved GPU's. I'm not saying the performance of Trinity isn't up to snuff, and I'm sure it's an improvement over llano, but how much of an improvement can't be evaluated based on two synthetic benchmarks.
 
tviceman said:
There needs to be a more thorough test than just 3dmark and pi. 3Dmark looks great on paper, even with extremely bandwidth starved GPU's. I'm not saying the performance of Trinity isn't up to snuff, and I'm sure it's an improvement over llano, but how much of an improvement can't be evaluated based on two synthetic benchmarks.
Click to expand...

Ofc. It's a very early leak.

I read somewhere Trinity is 2x shaders compared to Llano. There's room for a lot of improvement, and would make a viable gaming setup on the cheap playing most games at medium 1080p.
 
lOl_lol_lOl said:
If the cpu performance is any indication of the improvement 'Piledriver' brings to the table, then we might as well forget the upcoming desktop cpus all together.:thumbsdown:
Click to expand...

Though the CPU is only at 2ghz, its still a dismal looking outcome. At 3.7ghz my 2500K takes 7s to count to 1M pi, this one takes 23.
 
Silverforce11 said:
Ofc. It's a very early leak.

I read somewhere Trinity is 2x shaders compared to Llano. There's room for a lot of improvement, and would make a viable gaming setup on the cheap playing most games at medium 1080p.
Click to expand...

For you to know it is a very early leak means you know when it launches? Or is that public knowledge?
 
Smoblikat said:
Though the CPU is only at 2ghz, its still a dismal looking outcome. At 3.7ghz my 2500K takes 7s to count to 1M pi, this one takes 23.
Click to expand...

Compared to a FX-4100 at 3.6ghz that also pulls 23 seconds in a 1M pi, that doesn't look too bad.

http://www.mygarage.ro/attachments/...-uploadati-pozele-aici-superpi-stock-4100.png

^Stock

Next two are overclocked

http://hwbot.org/image/740578.jpg

http://extreme.pcgameshardware.de/m...re496554-img-amd-fx-series-bench-super-pi.jpg
 
To be fair you can't really compare AMD to Intel Superpi scores.

20s @ 2GHz is helluva impressive considering when it takes 3.6GHz PhII to reach there, and we know BD mostly cannot beat PhII core-for-core and clock-for-clock. Going by Superpi alone Trinity is 1.8x more IPC than PhII.
 
Silverforce11 said:
Ofc. It's a very early leak.

I read somewhere Trinity is 2x shaders compared to Llano. There's room for a lot of improvement, and would make a viable gaming setup on the cheap playing most games at medium 1080p.
Click to expand...

We'll see how actual gaming scenarios are affected by sharing system ram. That is going to be a big hurdle going forward for any integrated-graphics solution.
 
StrangerGuy said:
To be fair you can't really compare AMD to Intel Superpi scores.

20s @ 2GHz is helluva impressive considering when it takes 3.6GHz PhII to reach there, and we know BD mostly cannot beat PhII core-for-core and clock-for-clock. Going by Superpi alone Trinity is 1.8x more IPC than PhII.
Click to expand...


lol that would be awesome, but meh brain won't allow me to believe it. :\
 
Joseph F said:
Right, 2000MHz is the idle speed.
For those of you who didn't catch it, look below CPU-Z.
Click to expand...


Okay then... ouch indeed.

I was thinking it would take ~4GHz BD to get 23 seconds. 3.8 is close enough that I think single threaded performance will still be pretty dismal on Trinity then. While these are early benches, good luck convincing ANYONE that things are going to change dramatically on future revisions. Unless they're insane or dumb as a rock, they aren't going to fall for that two generations in a row.

While SuperPi doesn't represent more than a very narrow portion of the processor, it does expose a significant flaw in BD, the cache subsystem latency. IMHO we aren't going to see significantly better per thread BD performance until that latency is improved significantly... and that is the sort of thing that would show up in SuperPi.
 
Doesn't look like PD brings much to the table in terms of IPC improvements, but looks like they have managed to improve efficiency a ton to clock the Piledriver cores on Trinity so high. 3.8 base/4.2 turbo is pretty crazy for a 100W TDP APU, keep in mind that current Bulldozer quads with those frequencies are like 95W. Bodes well I hope for 125W Vishera later this year being able to achieve some pretty high stock clocks (low 4GHz base, mid/high 4GHz turbo is probably what they need to stay at least somewhat competitive with IB).

And GPU performance looks great of course.
 
Good job amd,you bring a package of near 8800gt performance and decent quad core performance together as one,lets hope for a low tdp number and low price.
 
I don't think this is indicative of Piledriver since this is the successor to Llano. I also don't see integrated graphics being capable of high quality 1080p gaming any time soon due to the lack of memory bandwidth.
 
Concillian said:
Okay then... ouch indeed.

I was thinking it would take ~4GHz BD to get 23 seconds. 3.8 is close enough that I think single threaded performance will still be pretty dismal on Trinity then. While these are early benches, good luck convincing ANYONE that things are going to change dramatically on future revisions. Unless they're insane or dumb as a rock, they aren't going to fall for that two generations in a row.

While SuperPi doesn't represent more than a very narrow portion of the processor, it does expose a significant flaw in BD, the cache subsystem latency. IMHO we aren't going to see significantly better per thread BD performance until that latency is improved significantly... and that is the sort of thing that would show up in SuperPi.
Click to expand...
FWIW my FX-8120 @ 4.4GHz gets 19.660s in Super Pi mod 1.5. 23.775s seems a bit slow if this thing turbos up to 4.2GHz. Lack of L3 must hurt Trinity in Super Pi, because all else being equal PD is supposed to have a bit higher IPC than Bulldozer (like 5-7% or so from what I've read I think).

edit: Here are the averages of 3 runs at various frequencies

3.8GHz - 22.520s
4.2GHz - 20.400s
4.4GHz - 19.491s
 
Last edited:
Smartazz said:
I don't think this is indicative of Piledriver since this is the successor to Llano. I also don't see integrated graphics being capable of high quality 1080p gaming any time soon due to the lack of memory bandwidth.
Click to expand...
Well, something to take into account is Trinity has a 2133mhz RAM controller... which if you have dual channel 2133mhz RAM it comes out to about the same memory bandwidth as you find in a Radeon HD 7750.
 
Vulpesveritas said:
Well, something to take into account is Trinity has a 2133mhz RAM controller... which if you have dual channel 2133mhz RAM it comes out to about the same memory bandwidth as you find in a Radeon HD 7750.
Click to expand...

I thought 2133MHz dual channel would be ~34GB/s effective memory bandwidth versus the 72GB/s on the 7750. Btw, did anyone notice that the chip is called the 5800K, perhaps they're trying to fool buyers into thinking this is better than the 3570K and 3770K?
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top