GTX 970 Owners Thread

Done some benches with different memory on these 970 cards.

1.970GTX G1 Gigabyte with Hynix Memory


2.970GTX G1 Gigabyte with Samsung Memory


1

Spoiler

Hynix
1404 Core
7008 Memory



Hynix
1537 Core
7600 Memory



Hynix
1593 Core
7600 Memory



Hynix Theif Benchmark
1593 Core
7600 Memory

2

Spoiler

Samsung
1404 Core
7008 Memory



Samsung
1537 Core
7600 Memory



Samsung
1593 Core
7600 Memory



Samsung Theif Benchmark
1593 Core
7600 Memory

 

Unless you're benchmarking, it really makes little difference.

My card has Hynix. If it had Samsung, I'd have returned the card already....

 

Why?

 

Holy fan RPM Batman! I couldn't stand that noise at all.

 

Thanks for sharing this information. I am completely out of the loop on these.

 

No offence but I don't believe you. I don't think even on benchmarks that there is -if any- such a difference on stock speeds...and seriously,you did some benchmarks but you own a AMD card..? I pass.

 

What is it that you don't believe?
Samsung uses stricter timing (lower latency settings) than Hynix.
Samsung is faster at the same frequency, but Hynix can be overclocked higher than Samsung cause of their looser timings.
This has been proven a few times.

 

People cannot own more than 1 brand of cards? That guy has benches all over this forum with amd and nvidia cards.

 

gerard wasn't happy with his 970's to he sold them and bought the 290's. Those benchmarks are from the 970's he used to have.

 

but he had 290's when they were new pretty sure

I benched a little bit sorry no gaming

bout all I can do on air clocks are alittle higher then gpuz reports 1580ish
not like GF'S but oh well


pretty sure I got 31k but I don't know where the ss is stuffs all over atm


vantage I have the ss somewhere
58,5k or so
http://www.3dmark.com/3dmv/5220029

I am just going to try some hard mods and get it -0c cold this week some time see what happens

 

100% CORRECT sometimes its best to have the Hynix under sub zero cooling:nerd:

 

Sorry man all I can say is that's what I got when I ran the benches and like TK guy said .I plastered lots of benches I did on this forum.

Yes this is 100% correct with the testing on the cards I did.The Hynix card overclocked like crazy not so much for the samsung card but it was still a great overclocker.

True I was not happy with 970 SLI performance(below normal).980 SLI was good.
Funny thing is,I got a return on the Hynix card and been using it for a couple days.Running it @1600/7600 and all is fine with single card.

cowie is correct also had
R9290 and r9 290x crossfire when they came out.

 

Ok,guys, sorry , my bad. However a "regular" OC'er won't be so absolute about this difference(seems kinda small),but yeah, a absolute/advnaced OC'er would prefer the best of the best, true.

 

Just ordered EVGA GTX970 SSC looking forward to overclocking it to see what it can do from what I have read should be around 1500 core with the extra voltage above the standard GTX970

 

Thanks, my cards has Hynix and this plus the information that followed
helps me to better understand the positives and negatives from using Samsung vs Hynix memory.

 

It's not all that simple though.
Those Hynix chips have quite some variation in how much they can be OC'ed before they show artifacts, that variation seems lower with Samsung chips.
It's a bit like overclocking several of the same CPU models but from different badges.
Either they go through the roof (clock way beyond what those Samsung chips are able too) or they don't.

 

Just because a chip can clock high, doesn't mean it's going to yield performance gains either. Every chip, whether it be a CPU, GPU or memory chip, has a break point.

 

Copy & paste from old post

I really do not think the overclock matters a great deal with some of these cards.Of course we all want the highest overclocks to feel you got a great deal but it is not always the fastest when gaming with higher overclocks.

EG: 970GTX G1 Gigabyte cards.When I tested them alone here what my results were in lets say valley benchmark.

1.970GTX G1 Gigabyte with samsung ram Clocked 1540/7800
Valley FPS 65

2.970GTX G1 Gigabyte with hynix ram Clocked 1605/7800
Valley FPS 63.5

The lower clocked card was faster,could be a few things but they still run good.A Higher clocking card is not always faster.

When in SLI overclocking results are also not that dramatic either with a few more Mhz.

EG:Some 970GTX Gigabyte G1 SLI on Metro Last Light(Original Game) a total difference of 3.7 FPS not really much when you think about it.
From default to almost 1600 on core overclock.

1392/7600 started at 1405 dropped to 1392 on core



1506/7600



1538/7600


1558/7600


1593/7600 Higher overclock less FPS

 

Hey guys,

It's safe use 1.27v constant in gigabyte g1 gaming?
my stock voltage is 1.21v

i never hit 70º in gpu core, but vrm is my problem...dont have monitoring in gpu-z!

 

check the g1 review here on guru. the vrms on the g1 run pretty cool compared to some other 970 iirc.