show me that a q6600 is no good now

[Shambles1980]

You're forcing your GPU to maintain 60fps, and to render 3x that. (Or so you say)

Having your CPU/GPU at 99% constantly in a game is not good for the health of the hardware.

no your missing the point of tripple buffering entierly..
your graphics card with no vsynk will send its renderd frames to the monitor when it renders them.. lets for arguments sake say your card is able to render a constant 180fps and your monitor is 60fps
in that instance your gpu is sending 3x more data to the monitor.
this can cause tearing because your monitor may well start to render frame 3 but as it started to do that frame 4 has been sent to it, so the top of the picture is renderd frame 3 and the bottom of it is rendered frame 4. if these 2 frames are a bit different to each other then you will see tearing, if they were almost identical then you wont see any tearing..

now what tripple buffer does is have 2 buffers and the front buffer so frame 1 goes to the monitor the card still renders the images like it did with no vsynk. but it goes to the buffer. "still no monitor refresh" the card renders another image and sends that to the buffer "still no refresh" the card renders another image and sends it to the monitor because it refreshed.. the 2 buffered images were never used but were rendered incase they needed to be..

now limited que also works very similar, but is more of a double buffer. where it saves the images that werent able to be displayed due to a refresh, and after a set time deletes them. this works almost exactly the same. but if the screen refreshes before the set ammount of time then it will display an image from the buffer. this image would be considered old as the gpu could have rendered one instead of it.
in reality though limited time que works very well.

double buffer is a bit like limited time que and tripple buffer but worse. as it only has 2 buffers then it sends image 1 to the monitor image 2 goes to the buffer and then it sits their till it can be displayed. the gpu was not able to buffer any more images and did not send out a new refreshed image and so the image from a double buffer is almost always old. but you can use double buffer with no v-sync which is kind of pointless because then you can still get tearing.

so the card isnt working any harder. its just preventing tearing with the bonus of not delaying images.

 

[TheoneandonlyMrK]

Since he is also benching default settings as well as his speculative own style its all good imho

 

[crazyeyesreaper]

meh Q6600 play Rome 2 Total War or Shogun 2 Total War maxed out with larger armies watch it crawl. 10-12 FPS in melee vs 29-30 of the newer Haswell chips lol

 

[Toothless]

Triple buffer in combination with v-sync doesn't mean it renders 60fps three times but to render max. 3 frames ahead of monitor sync..

THANK YOU. At least I can understand the way you explained it.

 

[OneMoar]

don't forget to get 800Mhz out of a 8150 you need a board with the VRM to handle that at 4.4 a FX chip is gonna be pushing 200W TDP to say nothing of the cooling required

 

[HalfAHertz]

That q6600 vs i7 4770k was pretty interesting. The q6600 held surprisingly well in gaming at 19 x 12 for a 7 year old cpu. However I think that the Cinebench test was pretty telling. The q6600 was getting only 0.69 points per core. Last time I checked the 8120 was around 0.90+ in that one. That's a 30% increase in single threaded cases and considering that the amd has twice the numbers of cores, things are looking good for the OP.

 

[Shambles1980]

its not the best board but it is listed as being able to oc. but only 4+1 so really not going to get that much out of an 8120 at all "like i said a few pages ago a 63xx would have been better all round but this is what i could affoard.
i will be adding heat sinks to the vrm's but i dont expect to see crazy over clocks.
as for the cpu its self. thats getting a xigmatek aegir sd128264 so thats covered, but the board being 4+1 i dont expect much from.
possibly around 4.2 maybe more maybe less if i add heat sinks and a 80mm fan for the vrm's
but some where in that general area i think is the target for 4+1

pretty sure i have some heat sinks here that will do the job nicely but i havent looked at the board properly yet hopefully it wont be difficult to keep the transistors cool

 

[HalfAHertz]

its not the best board but it is listed as being able to oc. but only 4+1 so really not going to get that much out of an 8120 at all "like i said a few pages ago a 63xx would have been better all round but this is what i could affoard.
i will be adding heat sinks to the vrm's but i dont expect to see crazy over clocks.
as for the cpu its self. thats getting a xigmatek aegir sd128264 so thats covered, but the board being 4+1 i dont expect much from.
possibly around 4.2 maybe more maybe less if i add heat sinks and a 80mm fan for the vrm's
but some where in that general area i think is the target for 4+1

pretty sure i have some heat sinks here that will do the job nicely but i havent looked at the board properly yet hopefully it wont be difficult to keep the transistors cool

I'd expect that your 8120 at 4.2 GHz will deliver similar performance to my 3330 at 3 GHz. I'm very happy with my 3330 and an hd7850 and I think you made a good choice.

 

[Shambles1980]

I'd expect that your 8120 at 4.2 GHz will deliver similar performance to my 3330 at 3 GHz. I'm very happy with my 3330 and an hd7850 and I think you made a good choice.

well my heat sink will take a wile to arrive (just checked estimated delivery date). But ram should arrive today...
the vrm's i can add a temp heat sink to for now. may just need to cut it down some more. but with stock cooling i wont try any oc just yet.
i do have my eye on some 6+2 boards which should let me unleash more of the power i dont mind spending a bit on that and recouping some of it from this board as it should work out ok.


edit

having looked at the board properly today it only has 4 chokes near the cpu which to me says its 3+1 but it does have high quality transistors. but if it really is 3.1 then this board is not for over clocking regardless of what people on the internet say. And i will need the transistors heat sink'd just to run at stock speeds without throttling (at least i think so) haven got that much experience with split power phases but my understanding of it is that if the power circuit is protected (this board is) it will throttle down and or turn off due to heat.
I may need to do some more research in the board but at least for now i will treat it as under powered and not an over clocker. as i would rather get back some of the money on it because i did not blow it up and get a reasonable 6+2

i find it strange that the internet says the board is 4+1 but from what i see on the board its a 3+1

 

[TheoneandonlyMrK]

The memory phase is likely separate from cpu vrms

 

[Shambles1980]

not so sure. just set it up went to set the thing as an oc set at 3.1 (should be stock) turned off the turbo stuff and amd versions of speed step thing, so it should sit at a constant 3.1 provided the vrm isnt getting hot.
started occt "cant really give it a proper run as occt dosent detect core temps so i can't trust it to stop before any damage"
and it dipped to 2.8 at one point, which means that it throttled due to the power phase being over worked. (thats at stock) so this has to be a 3+1..
Which will make the tests some what un fair. Like i said though i do have a 6+2 im trying to buy right now.
but there is no way i can oc with this board.
i added a 80mm fan over the transistors, which should help so it should not dip under normal gaming and stuff. but if it cant occt without throtteling then its 3+1
I didnt add a heat sink to the transistors though. the one i was going to use is a bit short.
i do have some memory heat sinks orderd which should fit, but if i get the 6+2 then it wont matter..

any way il run some tests with it now. but it is going to be hindered by the board's vrm.

 

[OneMoar]

afaik if its a G0 chip you should beable to get 4Ghz out of it with reasonable voltage a good binned chip will do 4.0 with 1.45V or less I knew a guy that hit 4.5 with 1.52 under water

 

[Shambles1980]

ok so thief bench test. same as before with the settings. the 8120 did a little better but not really enough to warrant much fan fare.
as usuall the game reports min fps as 0 because of delayed start but msi after burner repoted it as 14.5 (.6 slower than the q6600)
but through out the test the 8120 was able to be consistantly higer and ended up with 63.5 max (lower than the q6600)
and aproximatley 36 avarage (which is quite a bit better) given the lower max frames (but same 0 frames reported) to achieve this the 8120 had to be more consistant or with less low drops during the test.
the actual avarage reported by the bench was 33.9 (4.1fps better than then q6600 achived with higer max frames)

so it looks prety good for the 8120 so far for being more consistant..

the actuall game..
again it was reasonably close. with the 8120 being slighjtly more consistant at staying in the 60fps requested of it. but it too suffered some drops mostly in to the 50's but 1ce diped down to 41.5 (slightly lower than the q6600s lowest point) i must stress however that this was for one small second once, where as i could repeat the 44fps drop with the q6600 every time i ran down the open length of the court yard. the 8120 only did it once and managed to stay at 60.1 fps when i ran down the same path again a few times (possibly throtteling due to the vrm)

The only conclusion i can take from this is that with this board and cpu you would notice no reall difference gaming at the settings using this cpu at stock vs a over clocked q6600 at 3.0
Again i need to stress that this is a slighjtly unfair test due to the fact that this board is 3+1 and will occasionally throttle the cpu to 2.8 when under load.

on to the synthetic tests.
valley bench

min:
15.1 (dx11)
15.4(dx9)
14.6 (opengl)

Max:
74.1(dx11)
78.1(dx9)
66.1 (opengl)

average:
41.6 (dx11)
40.3(dx9)
32.5(opengl)

from here we again see that these 2 processors are again very close provided the q6600 is over clocked to 3.0. and the fx-8120 is locked to 3.1ghz and gets throttled down due to stupid 3+1 power phase.
(how often it gets throttled to 2.8 however i dont know. but if we assume it never does. and was always running at 3.1 with turbo mode off. then the 8120 is atleast on par with a q6600)
As i said earlier this test is very biast in favour of the q6600 due to the stupid 3+1 power phase board which I will be replacing..

any way 3d mark.

here the 8120 seems to do prety much exactly the same as the q6600 in the gpu test. which would imply that the gpu is the limiting factor in this test.
Suprizingly (atleast to me) the 8120 actually performs the physics tests better than the q6600 even when hindered.
and in the combined test its a more even playing feild but the 8120 at 3.1 seems to be able to keep the frames at a more consitsnat rate compared to the q6600..

the q6600 performs quite a bit better here in terms of actuall raw numbers but the graph seems to show the 8120 is more consistant even when hindered..
but in the physics test the 3.0 q6600 is outright better than the 8120 at 3.1..
Which is strange as in the previous physics test it was the other way arround. (again this could be due to throtteling)


so.. wit this set up and the 8120 Locked at 3.1ghz with occasional throtteling down to 2.8ghz "due to stupid mother board" and being unable to utilize built in features such as turbo mode.
the 8120 and 25% oc q6600 seem to be about even...

this is actually a better result for the 8120 than i expected when i found out i was stuck with a 3+1 phase board and it throttled down under load.
If the 8120 had been allowed to use its turbo mode then the q6600 would not have kept up at 3.0 but at 3.6.. I think it probably would have.

so in the end.. i have to say that a stock 8120 would be better than a 25% oc'd q6600. because it would be able to use its turbo mode and hit 4ghz.
unfortunatly i cannot test that out untill i get atleast a 6+2 board.

but these cpu's running at very similar clock speeds are really close in terms of power per mhz. I would have to say the q6600 has slightly better performance for its mhz, but the 8120 on the right board can surpass the q6600 in terms of over clocking.
and if we are looking at stock cooling all round with the correct boards for each cpu then the 8120 would be the winner simply due to the higher clock speeds.

Is it an upgrade?
Well if you did like me and simply swapped your q6600 for the 8120 then yes i guess it is..
if you would need to pay more for the 8120 setup than you could get for your Q6600.. Then probably not Unless you cannot over clock your q6600

hope to get the new board soon and update this with more valid results..

afaik if its a G0 chip you should beable to get 4Ghz out of it with reasonable voltage a good binned chip will do 4.0 with 1.45V or less I knew a guy that hit 4.5 with 1.52 under water

with the Go Q6600 its not really the cpu thats the issue (well temps are, but thats something you need to cater for) to oc the q6600 really high you need a board that supports fsb of 1600 native and then can over clock past that with ease..

a 1333 board can usually oc to about 1600 and then will really struggle to go above that. some of them cant even do that though. and if your board was a native 1066 then you would be aiming for 1333 or better as your over clock and may not even manage that.

the q6600 really is about 3.7 max on a decent after market air cooler imo. (if you want it totaly stable in occt) i doubt 1.45 would give 4ghz stable on a G0 regardles of the board. even at 3.75 my G0 wanted to draw 1.48v to be occt stable. but that was unsustainable on air.
4ghz is more than possible with a G0 on the right board with good liquid cooling "i dont mean closed loop system" but it would more than likley want to draw 1.5+ when under 100% load on all 4 cores. but i think you could do it at 1.55 or lower.

 

[OneMoar]

note for the money you spent on a 8150 you could have a H81 and a i5 4750
http://www.newegg.com/Product/Product.aspx?Item=N82E16813128649
http://www.newegg.com/Product/Product.aspx?Item=N82E16819116897
for gaming this setup will do better then the 8150
FX 8150 150.00USD
AM3+ motherboard with at least a 6+2 phase VRM to support the 8150 75.00
________
h81 motherboard 60.00
intel core i5 200.00
so the price diffrence is really about 35.00
ofc there are other differences but for the 35.00 usd in price you will save that within the first 3 months in power consumption alone if you can't use all 8 threads of the 8150 then its pointless to own it

 

[Shambles1980]

note for the money you spent on a 8150 you could have a H81 and a i5 4750
http://www.newegg.com/Product/Product.aspx?Item=N82E16813128649
http://www.newegg.com/Product/Product.aspx?Item=N82E16819116897
for gaming this setup will do better then the 8150
FX 8150 150.00USD
AM3+ motherboard with at least a 6+2 phase VRM to support the 8150 75.00
________
h81 motherboard 60.00
intel core i5 200.00
so the price diffrence is really about 35.00
ofc there are other differences but for the 35.00 usd in price you will save that within the first 3 months in power consumption alone if you can't use all 8 threads of the 8150 then its pointless to own it

i dont know how many times i have said this lol.
but the cheapest USED i5 cpu here in the uk costs more than i spent on this system.. that is without motherboard or ram. infact the cheapest used i5 that i saw was more than that i5 you linked to is new. But again that i5 you linked to is more than i spent on this system..(it also has no ram or motherboard)

if i was in the usa it would be different but i am not lol.
If it had been as low of difference in price as it is in the usa it would have been a no brainer. but given it was more than double the money here. then its verry dificult to justify.

 

[RCoon]

the 35.00 usd in price you will save that within the first 3 months in power consumption

Not likely, it would be closer to around $12 in a year in savings. The cost in power of both processors at stock clocks is a few dollars a year difference.

 

[OneMoar]

Not likely, it would be closer to around $12 in a year in savings. The cost in power of both processors at stock clocks is a few dollars a year difference.

not here it isn't lol
power bill is 289 a month here and I haven't even run the AC yet

 

[RCoon]

not here it isn't lol
power bill is 289 a month here and I haven't even run the AC yet

Good god.

 

[Shambles1980]

well i updated my first post in this thread to have the results all in one place which is easier to find.
Still dont think its a fair test but the 8120 did better than i expected given the limitations the board forced on it.

 

[Bucho]

Hi Shambles1980 and thanks for the test results.
I mentioned that earlier that the board you got really is outdated and rather a lower end class. The AMD 700 chipsets were the 740G, the 760G, the 780G and V, the 785G and the 790GX. These are the ones with integrated graphics. These were usually used in µATX boards and were not really ment for overclocking. The ones without an integrated GPU were the 770, 790X and 790FX. There you could find a few great boards for overclocking.
But there are way better and newer boards (with more features and they use less power) out there with the 990X/FX chipset or some with the 970. The ones with the 990 are kind of expensive but the 970 should be a lot cheaper. You might want to check the 970 models out before you buy because some of them are no good overclockers.

As for the results ... they don't surprise me. In some cases at the same frequency the 8120 should be better but only if the program is optimized for it or it's threads. I guess Battlefield 4 would show better results with the FX at these speeds.
And like you said the FX will probably reach higher OC frequencies but you need a good board and good cooling as well as a decent PSU for that.
Still with the Q6600 @ 3.6-3.7 it will be a close match with that FX-8120 also OCd to about 3.8GHz.

 

[Shambles1980]

if the q6600 was 3.7 and the 8120 was 3.8 I think the Q6600 would actaully be better for all round performance.
but to get a G0 q6600 to 3.7 stable you do need to spend money.

i am looking at used 990fx boards right now. hope to see one for a decent price. and trade out this board for that.

i am also looking for cheap used 1155 intel boards. the cpu's are still to expensive for now which means a lot of people arent trying to buy boards used because buying them seperatly is usually more expensive than a used bundle.
im hoping to buy really cheap and then later on get a 2nd/3rd gen i5 for a good price.
as my wife is american. i can search the american ebay market and hope to get a cheap one there. a cpu should fit in to one of the boxes they send our daughter every so often.

that project is a (if i find them) situation though..

This board i have right now..
Simply sux. its 3+1 phase. the internet told me it was 4+1 more reseach shows Only the 3.1 rev was 4+1 and the later revisons for some reason went to 3+1 the processor TDP was dropped from 125w to 95w so really the 8120 should only run at 2.8 and its only staying at stock speeds for as long as it is because of the 80mm fan i put in place,
on a diferent note some "hypothetical" questions made me change some settings and i can run this at 3.6ghz with no throtteling at stock voltages if i have it running as 6 cores.
so thats how il have it running for the time being. should help with single threaded apps untill i get a better board..

 

[Vario]

Try to find an i5 2400 (not 2400s) and a cheapo 1155 motherboard. You can even use the stock intel cooler. Its going to cost you less in the long run. Sell the 8120 and am3.

edit: i5-2300, 2310, 2320 can also be found cheap, theres a 2320 for $130 on ebay for example

 

[Shambles1980]

Try to find an i5 2400 (not 2400s) and a cheapo 1155 motherboard. You can even use the stock intel cooler. Its going to cost you less in the long run. Sell the 8120 and am3.

working on it but the 2400 is close enough to the same price as 2500k here that i may as well get the 2500k.(£10 difference)

any way the board was listed as 4+1 phase when its most definatly 3+1 so im seeing if they will do something about that.
I am trying to get a 6+2 board right now.
i have seen a really cheap board for about £25 that would let me put a 2/3gen i5 in it if i got one, and a really crappy celeron G something or other i could use mean while. but again the i5's are just still to expensive right now.

 

[Toothless]

Okay so I'm sitting here with an older desktop that has a Q6700 paired with GTS8600. It's the same exactly chip as the Q6600 but with a .22ghz higher clock. Runs every day-to-day thing smoothly and I'll be testing some games on it soon.

 

[Shambles1980]

the 8600 will be the issue with that setup. the best card i paired with my q6600 IMO was the 5770, it had a 3870 and a 4870 prior to that all of those cards however are much better than the 8600. and the 3870 was not really that great. i would have gone for a 6950 as i thought at the time it was about the limit of the q6600 and possibly wouldnt see the full potential i did not do that. but the 7850 i now have came allong at a really good price with prety much the same performance so i thought id try it. Seems like the 7850 is a bit to much card for the q6600 which has then led to all this lol.
i think i still have the 4870 here some where.