7950 Or 7970 £50 difference worth it?

Absorbentsolid · 15 Feb 2013 at 10:03

Hey so i was about to purchase the Gigabyte 7950 but i noticed that the Gigabyte 7970 is on sale the week for £50 more.Is it worth to take the £50 Leap? If so would i be getting a lot more performance for the extra £50 or is it not worth it?
Any help would be Great Guys Thanks a lot!

Space Monkey · 15 Feb 2013 at 10:11

Apparently the difference is about 5%, worth the extra? Only you can decide

Orangey · 15 Feb 2013 at 10:25

This might get me flamed but no it's not worth it unless you bench or game in eyefinity using crossfire.

tommybhoy · 15 Feb 2013 at 10:32

Difference is between 5-10% game dependent clock for clock(check my sig).

Is it worth it?

No, not really if it's for gaming, if it won't run on the 70, it won't run on the 50.

If you are into benchmarking, that's a different matter although the 7970 WF is voltage locked, the unlocked 7950 with voltage, could in theory oc higher than the 70 if you get a good one giving you almost the same performance.

Rusty0611 · 15 Feb 2013 at 10:46

It's 5% at the same clock speeds so not really no. If you're into benching then I'd say so.

Nuclear Fusion · 15 Feb 2013 at 11:17

I thought the 7970 WF was voltage locked but the 7950 WF wasn't? In that case, the 7950 is certainly the better option as it can be faster with a voltage increase than a locked 7970. If both are unlocked, it is still better value to get the 7950 as the 7970 is only around 5% faster at the same clocks.

kissenger · 15 Feb 2013 at 12:01

If you're talking about the volt locked 7970 WF, then don't even bother. It's not worth it. If the 7950 is unlocked then go for that.

However, the bang for your buck reputation of the 7950 is largely built on the time when 7950s were going for a little over £200. Now they are more expensive and the gap is much less, so an unlocked 7970 might be worth it, if you can get one for around £50 more. The performance increase is roughly 15%. It is around 5% if you run the 7950 at 7970 clocks, but that's not really a fair comparison because the 7970 tends to be able to run higher clocks than the 7970.

Rusty0611 · 15 Feb 2013 at 12:09

kissenger said:
but that's not really a fair comparison because the 7970 tends to be able to run higher clocks than the 7970.

Untrue.

I've had 4 7950s each of which did 1200 MHz comfortably, two of them did 1250-1275 MHz comfortably, the other two had minor core articfacting. By no means a total sample but that's a common misconception. The point is this is the same kind of ground as 7970s reach roughly.

The 7970s only on average go further due to the density of Lightnings/Matrix Platimums about as well as ones which have exotic cooling.

arc@css · 15 Feb 2013 at 12:10

7950 all the way.

kissenger · 15 Feb 2013 at 13:04

Rusty0611 said:
Untrue.

I've had 4 7950s each of which did 1200 MHz comfortably, two of them did 1250-1275 MHz comfortably, the other two had minor core articfacting. By no means a total sample but that's a common misconception. The point is this is the same kind of ground as 7970s reach roughly.

The 7970s only on average go further due to the density of Lightnings/Matrix Platimums about as well as ones which have exotic cooling.

I don't think that's true. I think for 7950s to get up to the upper reaches of 7970 overclocks you tend to need more voltage and better cooling.

tommybhoy · 15 Feb 2013 at 13:27

I have had 2 7970's and 2 7950's, the 70's oc'ed higher than the 50's, especially the memory.

I would still say for gaming, the 7950 is by far the logical choice, cheaper and will get you the same playability that the 7970 will.

Again, clock for clock is 5-10% game dependent, the 7970 stretches it legs more on the likes of Sleeping Dogs due to shader intensive settings, not the 5% ballpark figure that keeps cropping up in here(it's a bit pedantic from myself I know but just letting the op know).

Marine-RX179 · 15 Feb 2013 at 13:29

kissenger said:
I don't think that's true. I think for 7950s to get up to the upper reaches of 7970 overclocks you tend to need more voltage and better cooling.

Yea well, voltage locked 7970 the chances are they won't reach the "upper reaches of 7970" altogther anyway...a unlocked voltage 7950 is more likely to hit 1200MHz than a stock voltage 7970...

kissenger · 15 Feb 2013 at 13:40

Marine-RX179 said:
Yea well, voltage locked 7970 the chances are they won't reach the "upper reaches of 7970" altogther anyway...a unlocked voltage 7950 is more likely to hit 1200MHz than a stock voltage 7970...

Yes, that's obviously what I was implying in my earlier post. :/

Marine-RX179 · 15 Feb 2013 at 13:43

kissenger said:
Yes, that's obviously what I was implying in my earlier post. :/

Ah right. I thought you were saying 7970 would clock better, simply because it is a 7970, not a 7950

Rusty0611 · 15 Feb 2013 at 14:01

kissenger said:
I don't think that's true. I think for 7950s to get up to the upper reaches of 7970 overclocks you tend to need more voltage and better cooling.

It is true because that is my own experience

.

As I said, by no means a complete sample though. Voltage in line with what 7970 users are using to get the same clocks.

The memory is slightly worse on the 7950s I've found but a memory overclock from 1600 (average) to 1700-1800 doesn't make much difference to overall performance and definitely not enough to revise it from 5% to 5-10%.

The 7950s on the 7970 PCBs definitely equal a 7970 in clock speeds achieved on core and memory.

tommybhoy · 15 Feb 2013 at 14:26

Rusty0611 said:
The memory is slightly worse on the 7950s I've found but a memory overclock from 1600 (average) to 1700-1800 doesn't make much difference to overall performance and definitely not enough to revise it from 5% to 5-10%.

Because in shader intensive titles, it's ~10% extra performance on the 70's, vram/core has little to no bearing in this regard.

Rusty0611 said:
It is true because that is my own experience

I have tested them on like for like clocks, come back and tell me I'm wrong when you have actually compared the two yourself.

Rusty0611 · 15 Feb 2013 at 14:33

Errr I wasn't talking to you Tommy? So I'm not sure why you've responded to me when I was posting to somebody else like I was talking to you all along... OK. I was specifically talking about the overclockability of the 7950s not the 5-10% debate and just used this to frame my line of thinking.

Here is my take on the matter though as you seem to want to discuss with me: without proper benchmarks from you with screenshots/evidence, I'll stick to the ~5% clock for clock difference which has been widely reported.

Actually, I remember a while back I matched with my 680 a 7950 in Sleeping Dogs with Extreme AA and both were 5% slower than a 7970 at the same clocks (ish). It was way too far back to dig out though.

kissenger · 15 Feb 2013 at 14:36

Rusty0611 said:
It is true because that is my own experience .

As I said, by no means a complete sample though. Voltage in line with what 7970 users are using to get the same clocks.

The memory is slightly worse on the 7950s I've found but a memory overclock from 1600 (average) to 1700-1800 doesn't make much difference to overall performance and definitely not enough to revise it from 5% to 5-10%.

The 7950s on the 7970 PCBs definitely equal a 7970 in clock speeds achieved on core and memory.

I don't mean to doubt that the results you found are true. I'm sure your '50s did achieve those clock. I'm just saying that if you look at 7950 overclock threads from around the net you will find that results people tend to get in terms of max clocks lag slightly behind 7970. Tommy's results reflect this. Hexus also states this in their clock for clock comparison of the two cards.

Rusty0611 · 15 Feb 2013 at 14:45

Yeah it's difficult to quantify it without a proper large sample and taking anecdotal examples like that is troublesome for statistical analysis. We'll agree to disagree on the difference of 5%

. I'm sure OP is going to be happy whatever way he goes.

tommybhoy · 15 Feb 2013 at 14:49

I'm responding to the 5% comments for the sake of the op and others reading.

I did post evidence on someone else's thread a while back, like yourself, I'm not going to sit and look for them, I take you at face value when you know what your talking about(680v7950 with your graphs-I didn't call for screenshots/evidence) despite the widely reported figures that are out there.

If you can't do the same back, that's your prerogative, I only hope that it's not because you don't like saying a simple 'ok fair enough buddy'.

It might help others that want to know the same answer in the future, but it seems likely, in your eyes that your correct and I'm wrong despite never using/testing them yourself.

Why would I make it up(I know you never implied it, but it equates to me pulling numbers out the air) when I'm using both cards despite telling the op to save the money and get a 50 instead?

:confused:

Mind boggling to say the least.