AMD Radeon HD 7970 GHz Edition Review: Give Me Back That Crown!

Page 5 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.


The non K edition i5s, such as the i5-2400, tend to be unable to reach 4GHz. Since the 8120 can hit over 4.5GHz with all eight cores enabled, it stands to reason that it can go at least a little farther (with the same voltage) when it only has half of its cores enabled. The mod gives it somewhat more IPC than Phenom II. There is no way that a 3GHz to less than 4GHz i5 will beat a 5GHz or maybe even greater FX that has the same core count and slightly better performance than Phenom II at the same frequency.
 
Where is the proof... hmmmm

Wait, next you will tell me AMD GPU's have better IPC vs Nvidia.... and yeah overclocks really make that different?

Blaz, dunno who you are trying to fool, but we do not need wool pulling over our eyes..

Ohh wait, i do not want another 7850 fiasco... had to pull you up on that too...

How could the 6100 have worse IPC than a CPU with an identical architecture built on an identical stepping and process with the same mod? Besides, What kind of idiot would say that Nvidia has inferior performance per core to AMD? I'm not stupid, yet you always seem to think that I am.
 
the main issue is that both intel and Nvidia work as efficient, profit minded business men. AMD tend to work like innovators. no, I'm not tryint to make AMD look good with that statement, especially when AMD's innovations fail to deliver and give Intel a monopoly on the high-end market. but as is always the case with innovations, they may realize unforeseen potential in the future. for instance, AMD says "we think games may need 384-bit memory bandwidth a year or two from now, we should start testing that now", while Nvidia says "no, we don't need that now, we'll maximize performance on TODAY's games because that's how we'll be benchmarked, and just ask people to upgrade to the new card next year". both has their issues, especially when that 384-bit bandwidth becomes actually useful, scientists at AMD may have jumped on some other tech already...

Yes, I'm trying really hard to make a balanced statement here... sorry if it's confusing
 


you are 100% correct.
 


AMD? Innovators? alright then... AMD is the company that creates an architecture that is one of the best ever, Bulldozer, and implements it in such a horrible way that it isn't even able to compete without mods in most performance metrics that they must have crippled it intentionally. Let's look at BD. It was computer-designed instead of designed by expert engineers transistor-by transistor, supposedly a 20% increase in power consumption and die size with a 20% performance drop (that's a huge slightly over 40% drop in power efficiency alone), crap memory controllers, crap cache, soft-ended flip-flops, the list goes on. AMD might innovate new ideas, but they often do so in a very poor way.
 

verbalizer

Distinguished

see this is where the disclaimer should come in and should say something like:

disclaimer:
" these are my personal and biased comments on the performance of AMD and HD Radeon,
true techs will argue with me but do not take their word and 100's of reviews and benches.
I know what I'm talking about.... AMD to the death.! "
end disclaimer.

I'm not in the mood to link charts and prove this one wrong bro, I just don't have the energy.
I'm still in shock at your 'temporary loss of sanity' and that's being generous on how I feel.
 


Go ahead and show links. I'd like to see some.
 
Same architecture to Phenom II???

You have lost the plot pal...

Now I have no idea where you're going. I never said anything about Bulldozer having the same architecture as Phenom II. I said that with the mod, it gains better performance per Hz (which is not the same as IPC) than Phenom II. It's a very minor lead, but it's a lead. I also said that the FX-6100 and the FX-8120 have the same architecture.
 
AMD's catalyst driver has been criticized for a long time for being buggy and underperforming, so it would be fair give them credit when its due. Its great to hear good news from them, but I wouldn't be surprised when TG forums get crammed with "why doesn't my 7970 perform like the review?" topics.

Also, in the past companies have resorted to driver "shortcuts" that improved performance on certain conditions. Later on we found out they were cheating their way into better performance by secretly disabling some hardware-intensive tasks that had little impact on image quality (I remember NVIDIA doing this with Crysis).

In all, the top performance market has just gotten a little more competitive, which is great.

*And to the people flaming about Intel x AMD: Please get back on topic, or get a room.
 
after reading all the comments..wow.
i don't like the new card's power use figures....
i give credit where i think is due. imo amd built their first own (not like 6000 series) gpu architecture with gpu compute in mind as well as gaming and delivered. they had a full quarter lead over nvidia. they designed their gpus in a way that tsmc's new 28nm process could produce them more than nvidia's kepler. now with more mature process and amd's binning, amd can launch a better binned gpu. amd has kept their cards more available than nvidia's so far.
sure i like gtx 680 and 670's power efficiency and performance. but nvidia should not have resorted to blaming tsmc's 28 nm process.
edit: forgot one thing. nice showing with crippled vce and opencl support, amd. after monthes of wait, this is what you give users. shame!
 
I agree with other comments voicing a re-test for the 7870 and 7850 with these new drivers!

Nice improvements are always welcome, no matter the color of the team, remember that.

Great article as well, Mr Chris!

Cheers!
 



he didnt say that AMD was good at innovating. :pfff: :p
 

silverblue

Distinguished
Jul 22, 2009
1,199
4
19,285
Less of the arguments and downvoting, guys... we've been having a bit of an epidemic with the latter recently even with good posts and it's really not a surefire way of knowing something's not reading anymore.
 

Metroidam11

Honorable
Feb 19, 2012
25
0
10,530
Do you think Nvidia will release a GTX 680 ti? Personally, I do not think it is going to happen. But it would be cool if it did.
 

First, you know exactly what I'm saying... There's nothing 'reference' at all about the 1GHz AMD cards -- What's the difference in specs other than the Clocks?! AMD simply juiced the clocks - period.

There's no difference other than the fact you elected to not test/compare say the EVGA GPU's with the 'SOC' (labeled) vs 'GHz Edition' (labeled) GPU's. Anyone can stick an OEM (label) on anything - I'm not that naive.

IF you included OC results from comparing say a vanilla HD 7970 (OC) vs 'GHz Edition' HD 7970 then I'd buy into this a tad more, but only if you similarly did same on the GTX 600's. Your choice -- don't get upset if folks call you on it.

HD 7970 specs:
HD 7970 GHz Edition - http://www.amd.com/US/PRODUCTS/DESKTOP/GRAPHICS/7000/7970GHZ/Pages/radeon-7970GHz.aspx#3
HD 7970 Reference - http://www.amd.com/US/PRODUCTS/DESKTOP/GRAPHICS/7000/7970/Pages/radeon-7970.aspx#3
GCN Architecture
32 Compute Units (2048 Stream Processors)
128 Texture Units
128 Z/Stencil ROP Units
32 Color ROP Units
Dual geometry units

SAME

These 'versions' of HD 7970 aren't worth 1 cent more, for FREE use MSI Afterburner and wala (Magic) you have a GHz Edition - http://event.msi.com/vga/afterburner/download.htm

Juiced without even better cooling :pfff:
load%20temperature.png

load%20noise.png
 

markem

Honorable
May 1, 2012
37
0
10,530
PLS lets just except nvidia got pawned and stamped into the ground, where it rightfully belongs... whats going to be interesting is if these new partner cards will be setting more new world records and i cant wait for the AMD overclocked cards vs the nvidia overclocked cards

I put my money on the AMD (any bets?)
 


How did Nvidia get stomped into the ground and why do they belong there?
 
Status
Not open for further replies.