• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

4870 1GB Review up!

I managed to snag the 1Gb when OCuk were doing the for just under £190. Glad I did now. Its certainly a quick card. I had a 3870 before this. Only played Stalker Clear Sky (buggy **** BTW) and the difference is huge. I'm getting the same frame rates on max everything and 1920x1080 as I was with medium and 1280x720 on the 3870. The 4870 is a bit quieter too.
 
What are you talking about? The effect is greater at higher resolutions. The effect isnt simlar across all resolutions...Across resolutions the difference is 6.6%, 10.2% and 17.2% respectively.


sorry to sound blunt but i can read perfectly fine:)

james.miller said:
yes which clearly isnt a result of the extra ram if its happening across vastly different resolutions

at 1680x1050, that isnt a result of extra ram. its a result of faster ram.

but anyway, you are looking at the same skewed view of percentages as the review. it isnt 6.6% faster at 1680x1050 in conan, its a WHOPPING 2.4fps. 17.4% sounds like such an increase.....until you realise it's actually 3.8fps @ 2560x1600.


memory limited situations would show a more dramatic increase in performance just like grid does, which is the only game to do so.
 
Last edited:
I got the 4870 from ocuk this week for £188, replacing my 8800GT. Haven't had much time to fiddle but did get five mins on warhead. I have gained about 10fps, but in reality I couldn't really tell the difference if i turned r_displayinfo to 0.

A little disappointed atm, more testing to do!
 
yes which clearly isnt a result of the extra ram if its happening across vastly different resolutions. think about it.

on another note, no review should focus on percentages, it totally distorts the actual performance figures and is a real peeve of mine.

I take your point re. % quotes as opposed to real world fps gains, suppose it does paint a slightly different picture

Are there any posts/info anywhere about the memory timings on the new 1Gb cards, I can't remember seeing any, is it possible that future drivers could exploit such timings or the extra 512Mb memory too ?

Based on the £45-£50 price hike (RRP prices) then yeh the 1Gb is prob not worth the extra money today but if a good deal can be struck it's prob worth paying an extra £15-£20 maybe. If the Powercolor PCS variant with improved cooling that o/c to 830/4600 was sub £200 that'd be better
 
Read it again twice, have to disagree still. I think you have an issue with Anandtech and had made your mind up before reading the review, talk of memory timings is an assumption, bottom line is it's faster than the 512

He's not disputing that it's faster than the 512, merely that the increase in speed isn't due to the memory being increased from 512MB to 1GB as the increases are the same across all resolutions.

If the increased memory was beneficial then you'd only expect to see significant increases at the higher resolutions.
 
the 1gb card isn't worth it.

the %'s looks and sounds like theres a big different. but if u look at the fps theres really nothing in it,
 
Last edited:
He's not disputing that it's faster than the 512, merely that the increase in speed isn't due to the memory being increased from 512MB to 1GB as the increases are the same across all resolutions.

If the increased memory was beneficial then you'd only expect to see significant increases at the higher resolutions.

How do you know 1GB doesnt benefit the 1680*1050 resolution? You are making unfounded assumptions and anandtech isnt.
 
sorry to sound blunt but i can read perfectly fine:)



at 1680x1050, that isnt a result of extra ram. its a result of faster ram.

but anyway, you are looking at the same skewed view of percentages as the review. it isnt 6.6% faster at 1680x1050 in conan, its a WHOPPING 2.4fps. 17.4% sounds like such an increase.....until you realise it's actually 3.8fps @ 2560x1600.


memory limited situations would show a more dramatic increase in performance just like grid does, which is the only game to do so.

All I see are a load of assumptions. Also of course percentages are everything. name me one science that doesnt deal in relative terms? You're argument has weight when fps figures are in single digits and can be skewed a lot by normal variation, but in this case the numbers are large.
 
Last edited:
I got the 4870 from ocuk this week for £188, replacing my 8800GT. Haven't had much time to fiddle but did get five mins on warhead. I have gained about 10fps, but in reality I couldn't really tell the difference if i turned r_displayinfo to 0.

A little disappointed atm, more testing to do!
CPU limited?
 
How do you know 1GB doesnt benefit the 1680*1050 resolution? You are making unfounded assumptions and anandtech isnt.
I'm not making any assumptions myself, I'm merely explaining drunkenmaster's point, although on the face of it he does look to be correct as, whilst I can't be sure, I wouldn't expect the extra memory to have any appreciable effect at 1680x1050 resolution.
 
I'm not making any assumptions myself, I'm merely explaining drunkenmaster's point, although on the face of it he does look to be correct as, whilst I can't be sure, I wouldn't expect the extra memory to have any appreciable effect at 1680x1050 resolution.

I direct you to the 8800gt 256mb and 512mb review. Again such a drastic difference didnt always exist

http://www.anandtech.com/video/showdoc.aspx?i=3175&p=5

Look at the effect AA has as well.
 
I direct you to the 8800gt 256mb and 512mb review. Again such a drastic difference didnt always exist

http://www.anandtech.com/video/showdoc.aspx?i=3175&p=5

Look at the effect AA has as well.

You can't compare a jump from 256MB to 512MB with a jump form 512MB to 1GB. The jump form 256MB to 512MB was so drastic because 256MB is simply too low for titles at higher resolutions nowadays... whereas 512MB is enough for 95% of things, so you aren't going to see as much benefit at all going from 512MB to 1GB at the moment.
 
You can't compare a jump from 256MB to 512MB with a jump form 512MB to 1GB. The jump form 256MB to 512MB was so drastic because 256MB is simply too low for titles at higher resolutions nowadays... whereas 512MB is enough for 95% of things, so you aren't going to see as much benefit at all going from 512MB to 1GB at the moment.

But it shouldnt be level throughout. It should fall away at a resolution ith 256mb. Not be consistently worse across resolutions. The gap should widen as you go up resolutions. In some games this doesnt happen. This is the exact same argument. Doesnt matter on the RAM itself.

I'm just showing how assumptions shouldnt be made. You're making a second assumption to prove the first assumption correct. With no evidence to show either.
 
Q6600 at 3.4ghz so no I guess. I just shouldnt have listened to people posting on here saying its a massive diff going ftom 8800GT to this. It simply isnt in my eyes.
It's hard to not get caught up in other folks' excitement though! I did, but I went from a 6800GT to a 4870, like getting out of a Reliant Robin into a Veyron, something like 5fps -> 100+ fps in COD4. Can't believe the 6800 series was the bee's knees!
Wind up AA and AF - although you can't do like for like with your old card...
 
yes because games use more than 256mb, but current games don't use the full 512mb yet...

Again, that doesnt matter. The gap should WIDEN as the resolution increases since a higher resolution should give the 256mb card a harder job. In some games, the difference is consistent. This is the original argument. So is it timings again?

Consistency doesnt necessarily mean that it isnt due to the RAM. Not that the anandtech review showed consistency on this scale anyway.
 
Last edited:
But it shouldnt be level throughout. It should fall away at a resolution ith 256mb. Not be consistently worse across resolutions. The gap should widen as you go up resolutions. In some games this doesnt happen. This is the exact same argument. Doesnt matter on the RAM itself.

I'm just showing how assumptions shouldnt be made. You're making a second assumption to prove the first assumption correct. With no evidence to show either.

Hmm i'm not sure exactly what you mean as your english is a bit confusing in the way it's phrased... but I wasn't making random assumptions. Not all games perform similarly with regards to how they respond to memory amounts due to the often great differences in graphics, coding, genre etc. Benchmarks also never show hitching, which is something you will experience on a 256MB card playing a demanding game at higher resolutions. I have yet to see anyone with a 512MB 4870 mention they're getting any hitching.

As it stands with current games, 1680*1050 resolution is not high enough to make a 512MB 4870 struggle with VRAM, and going to 1GB won't net you any visible improvements.
 
As it stands with current games, 1680*1050 resolution is not high enough to make a 512MB 4870 struggle with VRAM, and going to 1GB won't net you any visible improvements.

So that isnt an assumption. Okay. Because you havent shown any evidence for this while I have the anandtech review.
 
Last edited:
I know because I play demanding games at that resolution with AA/AF and have recently gone from a 512MB card to a 1GB card. I didn't get any hitching with my 512MB card, and that was less efficient at using RAM than the 4870 is. The first mistake you're making is relying on one single review from a site that in the last 2 years has ha its reputation lowered considerably for providing poorly written reviews.

If a game not causing a card to run out of VRAM, then simply adding more will not make any difference.
 
Back
Top Bottom