Overclock.net › Forums › Graphics Cards › Graphics Cards - General › graphic card inquiries
New Posts  All Forums:Forum Nav:

graphic card inquiries - Page 3

post #21 of 25
Quote:
Originally Posted by Branjo View Post

I'd definitely go with a single card upgrade. I just put a second XFX HD6780 DD in my rig, but I got it for 75 bucks, so it was a very cheap gap filler until I can upgrade to a better single card with a lot of Vram. Plus there is heat issues to take into account when running an Sli / Xfire. I had to cut a hole in the side of my pretty Antec P182 and add a 120mm fan blowing on the cards, all I can say is a Dremel is the handiest thing in the world.
Oh yeah, top grade mods. Lol. I cut one in the back for a fan on back of CPU and vrms. Wasn't planning on selling it.
post #22 of 25
Quote:
Originally Posted by raghu78 View Post

so you realize the GTX 760 has the newer drivers. but you still question the correctness of the comparison. well fine. but lets not forget one thing when both the GTX 760 and GTX 770 cards are OC , the GTX 770 has 33% more shaders (1536 vs 1152) so it has more cuda cores to drive performance. also GTX 770 has more bandwidth to feed those cores when overclocked . the stock speeds are 224 Gb/s vs 192 Gb/s but you can see 7.8 - 8 ghz memory OC quite commonly on GTX 770 with OC. the GTX 770 card has the power phase circuitry to drive the memory at such high speeds .
i thought i mentioned before, not all drivers give an increase in performance and in some games the performance actually decreases. though i may at times wonder about alienbabel; there is always a stickied thread in the nvidia drivers section in the guru3d forum where i see folks roll back to a previous drive due to the performance decrease in their favorite game. i can also say from personal experience that when benching a few different driver releases i strap myself in for the roller coaster ride to follow. i'll spare you and any future reader the wall of text that will follow on how nvidia drivers affect OC speeds - its a harsher roller coaster ride - and finish by saying don't completely believe release notes.

i understand on paper the 760 has less than 30% cuda cores and ROPs with the same TMUs. but i am not really seeing that difference in other reviews/benchmarks. though if overclocking shows the difference more than at stock speeds . . . . . that is logical but i would be more comfortable with more than one set of data
Quote:
hardocp has other games too. metro with 1920 x 1080 very high settings and physx enabled shows 53.6 on GTX 760 OC vs 72.6 on GTX 770 OC. PhysX really stresses the GPU, so you can see the max diff of 35%. here the extra bandwidth cuda cores and clock speed seems to add to the gap. thumb.gif
fixed that for you wink.gif
unless i misunderstood you mentioning "bandwidth, which lead me to believe you were referring to Vram. physX strictly depends on cuda cores and speed. now in that scenario i can understand a 30%+ difference in performance when you throw physX in the mix.
Quote:
i never said that HD 7950(1.15 Ghz) = GTX 770(1.3 Ghz) in all games. i said in that particular game Hitman thats the case (or even sleeping dogs ). so in those games HD 7950(1.15 Ghz) is 25 - 30% faster than GTX 760(1.3 Ghz). in the other games like crysis 3 or farcry 3 the difference is 5 - 10%. in games like metro its more on par. so the GTX 760 in best cases can get close and in worst cases is far behind. on avg across many games HD 7950(1.15 Ghz) is atleast 10 - 15 % faster than GTX 760(1.3 ghz)
.

i apologize, i understand when comparing the 7950/770 you are talking about just a few games, hitman, sleeping dogs and i thought tomb raider was in the mix but maybe not . . but i certainly do not see an "atleast" 10%-15% difference with a 7950(1.15) vs 760(1.3) in the benches list on the [H] overclocking 760 review.
http://www.hardocp.com/article/2013/07/02/msi_n760_tf_2gd5oc_gtx_760_overclocking_review/5
CRYSIS 3 (1080 2txaa)
760 - 41 min, 108 max, 74.0 avg
7950 - 31 mi9n, 108 max, 75.5 avg
CRYSIS 3(1080 4x smaa)
760 -35 min, 85 max, 56.6 avg
7950- 32 min, 100 max, 60.4 avg
2% difference @ 2txaa and ~7% @ 4xsmaa
FAR CRY 3 (1080 2x msaa)
760 - 43 min, 59 max, 51 avg.
7950- 45 min, 64 max, 54.6 avg
FAR CRY 3 (1080 4x msaa)
760 - 36 min, 50 max, 42.8 avg
7950 - 37 min, 57 max, 45.9 avg
7% @ 2x msaa and 2x msaa
MetroLL (1080 AAA)
760 -42 min, 85 max, 64.4 avg
7950 - 45 min, 93 max, 66.2 avg
2% difference
TOMB RAIDER (1080fxaa)
760 - 25 min, 92 max, 53.2 avg
7950 - 36 min, 189 max, 68.4 avg
that is a big whopper @ 28% difference! but the avg difference is ~9% with tomb raider included and 5% without tomb raider in a direct comparison between the two cards. if you want we can have a bit of fun and say that hitman absolution would also be a 28% difference and add that to the average for a total of 11.5% difference. thow in sleeping dogs with an additional 28% and then the average shoots up to 13.625%. but seriously, are we not then looking at the majority of games that favor the AMD arch? toss in skyrim and i bet the average will start going back down.

so sorry looking at those figures i cannot agree that an OC 7950 (1.15) will on the average give a 10% to 15% performance increase over a 760 (1.3). the sweet spot for a 7950 to handily perform better than a 760 is ~1200. anything below a 10% and most definatly 5% in performance will highly unlikely be noticed in game play except at lower fps. (edit: and i didn't mention even less of a difference with min frames)
Quote:
clock for clock HD 7970 > HD 7950 >= GTX 770 on average across many games. so thats why a HD 7970(1150 - 1200 mhz) and GTX 770(1250 - 1300 mhz) trade blows depending on game. since on average the GTX 770 clocks 100 mhz higher than HD 7970 when both cards are overclocked, they are on par. but a golden HD 7970 at 1.3 ghz is faster across the board against a golden GTX 770 at 1.3 ghz in just about the majority of games. same for HD 7950 and GTX 670.

see, thats exactly where using a "control card" like the 770 to predict the performance difference between the 760 and 7950 can go all flonky. you show a very reasonable and logical comparison using the 770 but looking at what data i see for a direct comparison (in the above), it just doesn't exactly pan out that way.
Edited by looniam - 8/16/13 at 1:54pm
loon 3.2
(18 items)
 
  
CPUMotherboardGraphicsRAM
i7-3770K Asus P8Z77-V Pro EVGA 980TI SC+ 16Gb PNY ddr3 1866 
Hard DriveHard DriveHard DriveOptical Drive
PNY 1311 240Gb 1 TB Seagate 3 TB WD Blue DVD DVDRW+/- 
CoolingCoolingOSMonitor
EKWB P280 kit EK-VGA supremacy Win X LG 24MC57HQ-P 
KeyboardPowerCaseMouse
Ducky Zero [blues] EVGA SuperNova 750 G2 Stryker M [hammered and drilled] corsair M65 
AudioAudio
SB Recon3D Klipsch ProMedia 2.1  
  hide details  
Reply
loon 3.2
(18 items)
 
  
CPUMotherboardGraphicsRAM
i7-3770K Asus P8Z77-V Pro EVGA 980TI SC+ 16Gb PNY ddr3 1866 
Hard DriveHard DriveHard DriveOptical Drive
PNY 1311 240Gb 1 TB Seagate 3 TB WD Blue DVD DVDRW+/- 
CoolingCoolingOSMonitor
EKWB P280 kit EK-VGA supremacy Win X LG 24MC57HQ-P 
KeyboardPowerCaseMouse
Ducky Zero [blues] EVGA SuperNova 750 G2 Stryker M [hammered and drilled] corsair M65 
AudioAudio
SB Recon3D Klipsch ProMedia 2.1  
  hide details  
Reply
post #23 of 25
Quote:
Originally Posted by looniam View Post

fixed that for you wink.gif unless i misunderstood you mentioning "bandwidth, which lead me to believe you were referring to Vram. physX strictly depends on cuda cores and speed. now in that scenario i can understand a 30%+ difference in performance when you throw physX in the mix.

the more the work done, the more bandwidth is required. PhysX run on GPU definitely places additional bandwidth requirements.
Quote:
but seriously, are we not then looking at the majority of games that favor the AMD arch?[/I] toss in skyrim and i bet the average will start going back down. so sorry looking at those figures i cannot agree that an OC 7950 (1.15) will on the average give a 10% to 15% performance increase over a 760 (1.3). the sweet spot for a 7950 to handily perform better than a 760 is ~1200. anything below a 10% and most definatly 5% in performance will highly unlikely be noticed in game play except at lower fps. (edit: and i didn't mention even less of a difference with min frames)

In the last18 months the majority of big AAA titles were AMD GE. You cannot take that away as a factor. So definitely thats why AMD has the edge. and yes I do believe that HD 7950(1.15 Ghz) is atleast 10% or more faster on average. k. now let me prove it in a slightly different manner. HD 7970 Ghz(1050 mhz) = HD 7950(1125 Mhz) = GTX 770 which boosts upto 1130 mhz . this is well proven by benchmarks on the web and ocn. well guess what a GTX 760 can still not match a GTX 770 even when clocked at 1250+ mhz falling behind by 6 - 16% .

http://www.hardwarecanucks.com/forum/hardware-canucks-reviews/61844-nvidia-gtx-760-2gb-review-14.html
http://www.hardwarecanucks.com/forum/hardware-canucks-reviews/59265-his-hd-7970-iceq-x-hd-7950-iceq-x-review-14.html

throw in a lot more demanding games like metro (eg: crysis 3, farcry 3, tomb raider) and the avg diff would be 10 - 12% between a stock GTX 770 / HD 7950(1125 mhz) and a GTX 760(1.3 Ghz) .
Fragbox
(14 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i5 2400 DH67BL AMD Radeon HD 6900 Series XFX HD 6950 2GB 
RAMRAMHard DriveOptical Drive
Corsair Corsair DDR3 1333 Mhz 2 GB Corsair DDR3 1333 Mhz 2 GB Western Digital Caviar Green SONY DVD-RW AD-7260S 
OSMonitorKeyboardPower
Windows 7 Professional 32 bit BENQ G2420HD Logitech K200 Seasonic VX550 psu 
CaseMouse
ANTEC 200 V2 Logitech mouse 
  hide details  
Reply
Fragbox
(14 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i5 2400 DH67BL AMD Radeon HD 6900 Series XFX HD 6950 2GB 
RAMRAMHard DriveOptical Drive
Corsair Corsair DDR3 1333 Mhz 2 GB Corsair DDR3 1333 Mhz 2 GB Western Digital Caviar Green SONY DVD-RW AD-7260S 
OSMonitorKeyboardPower
Windows 7 Professional 32 bit BENQ G2420HD Logitech K200 Seasonic VX550 psu 
CaseMouse
ANTEC 200 V2 Logitech mouse 
  hide details  
Reply
post #24 of 25
Quote:
Originally Posted by raghu78 View Post


the more the work done, the more bandwidth is required. PhysX run on GPU definitely places additional bandwidth requirements.
as a whole yes, but when just looking at physX its memory requirements are not that much. i found a really good thread on [H] forums that test several different setups with several different physX cards:
http://hardforum.com/showthread.php?t=1684699
there is a specific post comparing physX cards; they used a single 680, SLI and 690 with (iirc) 650, 650ti, 550ti, 560 and 480 as a physX card and ran it through mafia 2, metro 2033 and batman: AC.:
http://hardforum.com/showpost.php?p=1038927066&postcount=27 just a quote from the thread! (Click to show)
Quote:
For the added heat, power consumption and etc that the GTX 480 brings I would say it isn't worth it for such minimal gains

I was hoping to see at least 10fps + gains on average at least.... but I didn't even get that

really the difference is so minimal that it appears once you get to the point of being good enough, the point of diminishing returns hits hard and steep. and taking a look at the 550ti with less than half the cuda cores and 80% less memory bandwidth than the 480 but is clocked 28.5% higher pretty much sums up what physX depends most on. i understand there is a difference adding a physX card than rendering physX on the primary gpu but all the difference is the calculations - that i do not believe gets loaded in memory - its done on the gpu with the bus not being involved.

Quote:
In the last18 months the majority of big AAA titles were AMD GE. You cannot take that away as a factor. So definitely thats why AMD has the edge. and yes I do believe that HD 7950(1.15 Ghz) is atleast 10% or more faster on average. k. now let me prove it in a slightly different manner. HD 7970 Ghz(1050 mhz) = HD 7950(1125 Mhz) = GTX 770 which boosts upto 1130 mhz . this is well proven by benchmarks on the web and ocn. well guess what a GTX 760 can still not match a GTX 770 even when clocked at 1250+ mhz falling behind by 6 - 16% .

http://www.hardwarecanucks.com/forum/hardware-canucks-reviews/61844-nvidia-gtx-760-2gb-review-14.html
http://www.hardwarecanucks.com/forum/hardware-canucks-reviews/59265-his-hd-7970-iceq-x-hd-7950-iceq-x-review-14.html

throw in a lot more demanding games like metro (eg: crysis 3, farcry 3, tomb raider) and the avg diff would be 10 - 12% between a stock GTX 770 / HD 7950(1125 mhz) and a GTX 760(1.3 Ghz) .

sure there are a number of GE games that have been released though looking how crysis 3 performs well on either a nvidia or amd gpu. established game engines switch between favoring one or the other but i don't see one developer not using what they have come accustomed to over the years because of GE titles being released. and slightly off topic ( or i ought to correctly say slightly even more off topic) i am not buying the amd hardware in consoles= pc games will run better on amd hardware theory. it would be silly to pigeonhole a game like that since PC games sales are still a large % of all sales with a large segment not amd hardware. but hey, what i think is logical, or even what is established doesn't always work out that way.

and speaking about how things work out in the real world; the best comparison between a 760 and a 7950 OC results would be have a direct comparison. using a third or fourth card(s) in relationships to either or both cards at best can give a theory. a theory isn't proven whether or not people accept it; tbh too many people are too readily accepting theories without proof - proof that can only be given with direct testing.

so my theroy with the data from a direct comparison is the 7950 will have a 5%-10% general performance advantage with a few games being a bit larger when both the 760 and 7950 are both overclocked (1.3/1.15). if you want to just look at hitman absolution, sleeping dog or tomb raider then it would be greater - but c'mon thats very biased.

if i understand correctly, your theory is the performance difference would be twice as much when using control cards in comparison.

i'm sorry but even trying to be open minded to other people's viewpoints; i just see too many variables involved with that; i'm just not comfortable with that method.
loon 3.2
(18 items)
 
  
CPUMotherboardGraphicsRAM
i7-3770K Asus P8Z77-V Pro EVGA 980TI SC+ 16Gb PNY ddr3 1866 
Hard DriveHard DriveHard DriveOptical Drive
PNY 1311 240Gb 1 TB Seagate 3 TB WD Blue DVD DVDRW+/- 
CoolingCoolingOSMonitor
EKWB P280 kit EK-VGA supremacy Win X LG 24MC57HQ-P 
KeyboardPowerCaseMouse
Ducky Zero [blues] EVGA SuperNova 750 G2 Stryker M [hammered and drilled] corsair M65 
AudioAudio
SB Recon3D Klipsch ProMedia 2.1  
  hide details  
Reply
loon 3.2
(18 items)
 
  
CPUMotherboardGraphicsRAM
i7-3770K Asus P8Z77-V Pro EVGA 980TI SC+ 16Gb PNY ddr3 1866 
Hard DriveHard DriveHard DriveOptical Drive
PNY 1311 240Gb 1 TB Seagate 3 TB WD Blue DVD DVDRW+/- 
CoolingCoolingOSMonitor
EKWB P280 kit EK-VGA supremacy Win X LG 24MC57HQ-P 
KeyboardPowerCaseMouse
Ducky Zero [blues] EVGA SuperNova 750 G2 Stryker M [hammered and drilled] corsair M65 
AudioAudio
SB Recon3D Klipsch ProMedia 2.1  
  hide details  
Reply
post #25 of 25
Quote:
Originally Posted by looniam View Post

and speaking about how things work out in the real world; the best comparison between a 760 and a 7950 OC results would be have a direct comparison. using a third or fourth card(s) in relationships to either or both cards at best can give a theory. a theory isn't proven whether or not people accept it; tbh too many people are too readily accepting theories without proof - proof that can only be given with direct testing. i'm sorry but even trying to be open minded to other people's viewpoints; i just see too many variables involved with that; i'm just not comfortable with that method.

k. you can believe GTX 760(1.3 ghz) is 5 - 10% slower than HD 7950(1.15 ghz) while I stick to my belief the diff is 10 - 12% or even more. frankly crysis 3 and farcry 3 are games where Nvidia does well and there itself the gap according to hardocp review was 7% with 4X AA. so I can right away say you are over optimistic about a 5% avg gap. the gap in GE titles is so massive that the avg is definitely above 10%.
Fragbox
(14 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i5 2400 DH67BL AMD Radeon HD 6900 Series XFX HD 6950 2GB 
RAMRAMHard DriveOptical Drive
Corsair Corsair DDR3 1333 Mhz 2 GB Corsair DDR3 1333 Mhz 2 GB Western Digital Caviar Green SONY DVD-RW AD-7260S 
OSMonitorKeyboardPower
Windows 7 Professional 32 bit BENQ G2420HD Logitech K200 Seasonic VX550 psu 
CaseMouse
ANTEC 200 V2 Logitech mouse 
  hide details  
Reply
Fragbox
(14 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i5 2400 DH67BL AMD Radeon HD 6900 Series XFX HD 6950 2GB 
RAMRAMHard DriveOptical Drive
Corsair Corsair DDR3 1333 Mhz 2 GB Corsair DDR3 1333 Mhz 2 GB Western Digital Caviar Green SONY DVD-RW AD-7260S 
OSMonitorKeyboardPower
Windows 7 Professional 32 bit BENQ G2420HD Logitech K200 Seasonic VX550 psu 
CaseMouse
ANTEC 200 V2 Logitech mouse 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Graphics Cards - General
Overclock.net › Forums › Graphics Cards › Graphics Cards - General › graphic card inquiries