• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD FX6100 Reviewed!

Good review, hope to see more :)

For me personally BD was a big letdown as i had high hopes..it probably means im going eat humble pie and go of Intel/ASUS next upgrade even after having massive problems with both. Intel seem to be 2 gens ahead at the moment the BD concept looks good on paper for me the power draw was the killer and seemed to point at production problems rather than design,.

Software's still an issue too, the core usage jumps around all over the place at the moment so I'd hazard a guess that the scheduler fix hasn't properly fixed it.
 
Software's still an issue too, the core usage jumps around all over the place at the moment so I'd hazard a guess that the scheduler fix hasn't properly fixed it.

noticed the same thing on mine, randomly spiking from core to core, no obvious pattern just totally random, unless you create profiles in Overdrive and tell it to run certain programs (Supreme Commander for example) on certain cores, that helps smoothness an absolute ton in that game I found, performance felt tons better even though Turbo Core is active, and should therefore be boosting frequency in that sort of application that doesn't use eight threads, but the random 'core jumping' somewhat makes Turbo Core not work all that well.

have you also noticed the peculiar non-linear scaling you get with frequency sometimes, where the performance % gained is higher than the % frequency increase? cache problem surely! :confused:

Edit: also not to go off topic too much, did you know you could fit 69 'Bobcat' cores onto a Bulldozer die, 320MM2 for Bulldozer, 4.9MM2 for Bobcat, lets assume that the number came down to let us say 48 bobcat cores, would a server processor with 48 of those cores not do more work with less power consumption than a one with eight Bulldozer cores, since isn't Bobcat faster, more power efficient and faster clock for clock vs. something like Atom, and since the server space tends to be very threaded, all those cores might pay dividends! even though I own a Bulldozer think Bobcat is AMD's true success story in recent times, that or Llano.
 
Last edited:
noticed the same thing on mine, randomly spiking from core to core, no obvious pattern just totally random, unless you create profiles in Overdrive and tell it to run certain programs (Supreme Commander for example) on certain cores, that helps smoothness an absolute ton in that game I found, performance felt tons better even though Turbo Core is active, and should therefore be boosting frequency in that sort of application that doesn't use eight threads, but the random 'core jumping' somewhat makes Turbo Core not work all that well.

have you also noticed the peculiar non-linear scaling you get with frequency sometimes, where the performance % gained is higher than the % frequency increase? cache problem surely! :confused:

Edit: also not to go off topic too much, did you know you could fit 69 'Bobcat' cores onto a Bulldozer die, 320MM2 for Bulldozer, 4.9MM2 for Bobcat, lets assume that the number came down to let us say 48 bobcat cores, would a server processor with 48 of those cores not do more work with less power consumption than a one with eight Bulldozer cores, since isn't Bobcat faster, more power efficient and faster clock for clock vs. something like Atom, and since the server space tends to be very threaded, all those cores might pay dividends! even though I own a Bulldozer think Bobcat is AMD's true success story in recent times, that or Llano.

There's a few instances I've seen a higher performance gain than the frequency increase. Some AIDA64 benchmarks and a couple of CPU limited gain (eg, DoW2 gave me a 55% increase from a 45% OC)

I might have to try the profiles out to see if that helps, if the scheduler gets properly sorted out then it should give a fair chunk of extra performance as well.

I wonder if Windows 8 will help or if the scheduler patch for Windows 7 that's been given out is the best that they can do?
 
the interesting thing is, software is more important than people give it credit for, people always claim hardware is the most relevant of the two, however so far I am tending to lean toward what is running being more important than what is running it in the current computer world.

example of this, did a fresh install of my OS (Vista 64-bit Premium) a while back, ran Cinebench on this processor running at 8150 speed (3.6GHZ), in the multi-core test it scored an alarmingly low score, which improved with the installation of Service Pack 1, which then improved some more with Service Pack 2. seems to be the older the software, the worse Bulldozer performs, where-as the newer the software, the better it performs.

and the more than 100% scaling has to be down to cache on the chip, has to be down to latency and with higher clocks, latency decreases (according to Sandra, and AIDA64) which means the cores spend less time sitting around waiting for data and more time actually executing the data.

do honestly believe that Bulldozer has more potential instructions per cycle than 'Stars' does but gets hampered by its cache, keeping those 'reasonably powerful' integer clusters from doing their work as efficiently as they can. compared to Sandy Bridge in some situations a Bulldozer core has to wait almost twice as long to receive instructions from its memory, so at the end of the day the dodgy performance isn't likely down to the front-end, the stream-lined 'cores', the shared resources but more down to the chips on-board memory subsystems trickling the cores data. also the problem of applications not making good use of its instruction sets doesn't really help matters much.

whilst I think a better scheduler will help in some respects (emulating, hopefully setting custom affinity to maximise the performance of each core in demanding applications rather than sharing) it won't make a dramatic effect. think the dramatic difference can only come from the drawing board itself, either A) somehow dramatically reduce the cache latency or B) dramatically bring down power consumption, so the cache works the way it was supposed to, at higher frequencies, B looks the most likely at the moment with the word on the street about Trinity, running at considerably higher clock speeds than Llano but consuming less power, at the same time can only point toward something changing at the fabrication level.
 
also why compare a BD to a I3?

a BD 4100 should be compared with a I3

a BD 6100 should be compared with a I5

a BD 81x0 should be compared with a I7

and i don't mean about price..
 
Last edited:
Why you compare Overclocked bulldozer to not overclocked i3 ???

Because, as stated in the OP, they're about the same price. You can't overclock an i3 so I can't compare to an overclocked i3.

also why compare a BD to a I3?
a BD 4100 should be compared with a I3
a BD 6100 should be compared with a I5
a BD 81x0 should be compared with a I7
and i don't mean about price..

Why would I compare it to an i5? The i5 is almost twice the price that I paid. When you're looking at what things should and shouldn't be compared to it's price bracket should be pretty much the only thing you should be considering :confused:
 
and i don't mean about price..
face_palm_by_Draculasaurus.gif
 
also why compare a BD to a I3?

a BD 4100 should be compared with a I3

a BD 6100 should be compared with a I5

a BD 81x0 should be compared with a I7

and i don't mean about price..

That doesn't make any sense. If you considering buying a Bulldozer 6100 based system why would I even be looking at Intel i5's?
 
I dont even acknowledge this POS as a 6 core CPU, it's nothing more than a 3 Core with AMD's version of hyperthreading.

Posts like this anoy me.

You have 3 modules each with 2 cores.

core2duo processors were one module with 2 cores.


I look at Bulldozer being the poor mans operton, however ive had a full 3 hour video encoding on 2 cores on my BD and still quite happily play BF3 with no slow down at all due to the way you can allocate the BD resources out. Never been able to do that on my 2600k
 
no no no and no, look at the slides where AMD positioned their product, it is not positioned against the 2600K and the likes, the 8*** series is intended to be marketed against the 25** series Intel products. don't compare processors that aren't in the same performance bracket, or your wasting your time.

same thing goes for comparing Bulldozer to Sandy Bridge clock for clock, its a comparison that shouldn't ever be done, they have never been the same type of architecture so any clock for clock comparison is pointless, unless your an Intel purist who loves the go on about instructions per clock.

Edit: also Bulldozer regardless of how much people keep pussy footing around it is a 'quad-core' with two integer engines in each 'core', so it should be regarded like AMD answer to Hyper Threading rather than a true 'dual-core', though CMT (Bulldozer) scales considerably better than Hyper Threading.
 
Seriously, Anyone that buy's a Dozer shouldn't be allowed near a computer.

No matter how many times people try to justify them either with price or overclock in mind they're still rubbish and get trashed by much cheaper Intel chips.

It may have beaten the i3 when overclocked but the i3 can also overclock, An overclocked i3 would smoke this chips wimpy ass...
 
no no no and no, look at the slides where AMD positioned their product, it is not positioned against the 2600K and the likes, the 8*** series is intended to be marketed against the 25** series Intel products. don't compare processors that aren't in the same performance bracket, or your wasting your time.

same thing goes for comparing Bulldozer to Sandy Bridge clock for clock, its a comparison that shouldn't ever be done, they have never been the same type of architecture so any clock for clock comparison is pointless, unless your an Intel purist who loves the go on about instructions per clock.

Instructions per-clock is the most important factor when it comes to selecting CPU's for gaming...

And as gamers we strive to get the best frame rates we can get for our money and no matter what price bracket you look at Intel is way in front of AMD.

AMD fan boys who were stupid enough to buy faildozer just try to dismiss the importance of per clock performance merely as a form of damage control.

I think it' safe to assume that if we round up every decent game release for say, the last 5 years most of them would still be only single or dual threaded and in that case AMD fan boys argument of IPC doesn't matter becomes laughable.
 
Edit: also not to go off topic too much, did you know you could fit 69 'Bobcat' cores onto a Bulldozer die, 320MM2 for Bulldozer, 4.9MM2 for Bobcat, lets assume that the number came down to let us say 48 bobcat cores, would a server processor with 48 of those cores not do more work with less power consumption than a one with eight Bulldozer cores, since isn't Bobcat faster, more power efficient and faster clock for clock vs. something like Atom, and since the server space tends to be very threaded, all those cores might pay dividends! even though I own a Bulldozer think Bobcat is AMD's true success story in recent times, that or Llano.

You fancy designing the pin-out for that? :eek:

Llano is fantastic btw, played with a A8 3870k system recently, had high hopes to begin with and was blown away :cool: GPUs sub £100 are well and truly obsolete
 
Seriously, Anyone that buy's a Dozer shouldn't be allowed near a computer.

No matter how many times people try to justify them either with price or overclock in mind they're still rubbish and get trashed by much cheaper Intel chips.

It may have beaten the i3 when overclocked but the i3 can also overclock, An overclocked i3 would smoke this chips wimpy ass...

Really ? This is what you think ? Thats quite an harsh statement to make. SIngle threaded applications fair enough arnt that great but for multi threaded applications it has been shown to surpass the i3 and sometimes is on par level with the 2500k, and since its priced very similar to the i3 i feel that price to performance ratio is fantastic. sure its not what everyone thought it was going to be but I use the 6100 in my second rig with a 6870 and game such as BF3 play at full 1080p at high/ultra settings
 
Really ? This is what you think ? Thats quite an harsh statement to make. SIngle threaded applications fair enough arnt that great but for multi threaded applications it has been shown to surpass the i3 and sometimes is on par level with the 2500k, and since its priced very similar to the i3 i feel that price to performance ratio is fantastic. sure its not what everyone thought it was going to be but I use the 6100 in my second rig with a 6870 and game such as BF3 play at full 1080p at high/ultra settings

I had a 4.8Ghz Phenom 2 x6 1075T with SLI 470's and it bottlenecked to ****...
 
Back
Top Bottom