Interesting 470 overclocked article on tweaktown

[Robes]

Why do people stick up for specific companies!? It's so weird! I bet they've never brought a toaster and said "Morphy Richards is the best, Russell Hobbs sucks!".

So why do it here? Fact is the new Nvdia cards are below par, their 6 months late, too hot, too expensive and too power hungry. However when Nvidia make another good card I'll buy it. Call a turd a turd for hevens sake.

 

[kylew]

Why do people stick up for specific companies!? It's so weird! I bet they've never brought a toaster and said "Morphy Richards is the best, Russell Hobbs sucks!".

So why do it here? Fact is the new Nvdia cards are below par, their 6 months late, too hot, too expensive and too power hungry. However when Nvidia make another good card I'll buy it. Call a turd a turd for hevens sake.

 

[R088ieS86]

Despite its downsides its good for people who have spent out to see that they can get a little extra out of the GTX470.

The GTX480 ISN'T the fastest GPU ever.

is the 480 not the fastest single GPU atm? (minus all of its downsides)
don't suppose anyone has come across any 470 SLI benches yet?

 

[kylew]

Despite its downsides its good for people who have spent out to see that they can get a little extra out of the GTX470.


is the 480 not the fastest single GPU atm? (minus all of its downsides)
don't suppose anyone has come across any 470 SLI benches yet?

You have to look at it this way, practically all 5850s and 5870s can overclock to 1000Mhz core, the GTX480 can't overclock enough to give more performance than a 5850 or 5870 at 1Ghz core.

Therefore it's not the fastest GPU because it's not capable of beating the competing GPU.

The clockspeed limit of Cypress GPUs is in excess of 1Ghz, that's the limitation of the GPU itself, the clockspeed of GF100s seems to be not more over 750Mhz which isn't a big enough speed boost for it to beat a Cypress GPU.

 

[Kribby2006]

Why do people stick up for specific companies!? It's so weird! I bet they've never brought a toaster and said "Morphy Richards is the best, Russell Hobbs sucks!".

So why do it here? Fact is the new Nvdia cards are below par, their 6 months late, too hot, too expensive and too power hungry. However when Nvidia make another good card I'll buy it. Call a turd a turd for hevens sake.

Lol this post wont go down well with the Russell Hobbs fanboys you know.

 

[bru]

You have to look at it this way, practically all 5850s and 5870s can overclock to 1000Mhz core, the GTX480 can't overclock enough to give more performance than a 5850 or 5870 at 1Ghz core.

Therefore it's not the fastest GPU because it's not capable of beating the competing GPU.

The clockspeed limit of Cypress GPUs is in excess of 1Ghz, that's the limitation of the GPU itself, the clockspeed of GF100s seems to be not more over 750Mhz which isn't a big enough speed boost for it to beat a Cypress GPU.

oh for gods sake, don't talk such rubbish, yes so gpu X can overclock more than gpu Y, unfortunately that doesn't make it the fastest, the gpu Y crowd could strike back with the fact that their gpu's shaders actually run at 1401MHz, but then you would stile back with under liquid nitrogen the X gpu reached 1600+ etc etc etc.

bottom line...... the gtx480 is currently the fastest single gpu on the market......this doesn't make it the best or the cheapest or the coolest or quietest or best bang for buck or anything else. but it is the fastest.

now of course we all know that their is a faster card on sale from the ATI camp but that has two gpu's so the original statement that its the fastest gpu is correct.

 

[Mr Krugga]

If AMD released an overclocked version of 5870, called it 5890, give it 2GB of RAM, give it 3 DVI/HDMI outputs for Eyefinity, redesigned cooling so it can run at higher clocks with low temps stable, make it 1GHz on GPU and 5GHz on RAM, we have the fastest single GPU in the world which runs cooler, performs better and costs the same as 480GTX. And it's totally doable now.

 

[bru]

yes it is

 

[CAPTAINBUMFLUFF]

Most likely will happen thats why I am holding back buying a 5850 now as they say a new revision is in the works

 

[kylew]

oh for gods sake, don't talk such rubbish, yes so gpu X can overclock more than gpu Y, unfortunately that doesn't make it the fastest, the gpu Y crowd could strike back with the fact that their gpu's shaders actually run at 1401MHz, but then you would stile back with under liquid nitrogen the X gpu reached 1600+ etc etc etc.

bottom line...... the gtx480 is currently the fastest single gpu on the market......this doesn't make it the best or the cheapest or the coolest or quietest or best bang for buck or anything else. but it is the fastest.

now of course we all know that their is a faster card on sale from the ATI camp but that has two gpu's so the original statement that its the fastest gpu is correct.

Wow, what passion.

Either way you look at it, you're talking about the capabilities of a GPU.

Would you say 1Ghz RAM is outright faster than 800Mhz RAM when the 800Mhz RAM is rated for 1.3Ghz and the 1Ghz RAM is rated at 1.05Ghz?

Anyone can raise the 800Mhz RAM to 1.3Ghz, it's the faster RAM, just because it's running at a lower clockspeed by default means nothing.

The GTX480 CAN NOT overclock high enough to beat an overclocked 5870, and considering all 5870s seem to overclock to 1Ghz, it means it's the faster GPU.


TL;DR

All it takes is for AMD to release a normal 5870 with a new bios at 1000 core/1300 memory and appropriate volatages, and it's the fastest GPU, they don't need to make a new GPU, they can just release new cards with the same hardware and a new bios. This is what makes it the fastest GPU, if the GTX480 was the faster GPU, it would would be able to overclock high enough to top an overclocked 5870.

Why does it take AMD to officially release a card with a modded bios for you to consider it to be the fastest GPU?

 

[And]

 

[Jayce85]

The GTX480 ISN'T the fastest GPU ever, it's more expensive than the fastest by around 50%, it's a huge fail.



DX11? Eyefinity? More performance?

I'd rather a 5850 over crossfired 4890s.

What is 50% cheaper but much faster than the GTX480 exactly?

As for Dx11, I'm pretty sure that a lot of its features are to be included for the Dx10 API.

Eyefinity? Surely a 50" Full HDTV would look nicer than putting 3 smaller PC monitors side by side?

 

[kylew]

What is 50% cheaper but much faster than the GTX480 exactly?

I said 50% more expensive not 50% cheaper, there's a big difference. I also didn't say much faster either. I was talking about a 5870, a GTX480 is 50% more money yet a HD5870 has the faster GPU as I stated in my last post:

All it takes is for AMD to release a normal 5870 with a new bios at 1000 core/1300 memory and appropriate volatages, and it's the fastest GPU, they don't need to make a new GPU, they can just release new cards with the same hardware and a new bios. This is what makes it the fastest GPU, if the GTX480 was the faster GPU, it would would be able to overclock high enough to top an overclocked 5870.

As for Dx11, I'm pretty sure that a lot of its features are to be included for the Dx10 API.

If DX11 features were to be included in DX10, then that would make DX11 pointless.

Eyefinity? Surely a 50" Full HDTV would look nicer than putting 3 smaller PC monitors side by side?

A 50" HDTV is comparatively low res compared to an Eyefinity set up firstly, I have 3x 24" monitors, that's nearly 7 mega pixels, a single 1080p TV is just over 2 megapixels.

Aspect ratio, triple monitors offers far more peripheral vision than a 16:9 TV, you will see a lot more of the game with 3 monitors than you will with a 50" TV.

My 3x 2408s have an aspect ratio of 48:10, that's far more information on screen than a 16:9 aspect ratio.

In my opinion a large TV would be far too big for PC usage, you can sit much closer to monitors than you can with TVs without worrying about eyestrain.

 

[Lightnix]

If DX11 features were to be included in DX10, then that would make DX11 pointless.

Well, strictly speaking DX11 did introduce the HLSL compute shader (at a lower, but still usable level of functionality) for DirectX 10 and above cards with shared memory support.

 

[kylew]

Well, strictly speaking DX11 did introduce the HLSL compute shader (at a lower, but still usable level of functionality) for DirectX 10 and above cards with shared memory support.

I am aware that some DX11 features are compatible, especially DX10.1 features, however the way Jayce said it came across as if he was saying "DX11 features are being added in to the DX10 API".

Which we know isn't going to happen as it'd make DX11 less worthwhile.

 

[Lightnix]

I am aware that some DX11 features are compatible, especially DX10.1 features, however the way Jayce said it came across as if he was saying "DX11 features are being added in to the DX10 API".

Which we know isn't going to happen as it'd make DX11 less worthwhile.

I see what you mean actually, but then I was just assuming the other guy got his terminology incorrect (they'd never change the DirectX 10 API specification itself at this point, it's way, way past done deal), so as to mean they might add things to future DirectX APIs which expose previously unexposed functionality supported by older DirectX 10 supporting graphics cards.

 

[Mr Krugga]

On the other hand (to be a little objective in the ongoing discussion) nVidia may have released a very capable GPU indeed. Let's just assume the prices of 480s drop down to £400 at least and we start seeing these GPUs with watercooled blocks just below £500, we might have the fastest single GPU ever which does not have all the cons of the current referenced versions of 480 - heat output of a volcano, noise of a jet and power draw of... I have to stop it, bills will remain the same or likely be higher when cards get some decent overclocks

I would personally love to see these cards in action and am sure there would be some enthusiasts willing to pay a grand for two of these cards for SLI setup. They would surely do well in DirectX 11 intensive apps in the top high-end systems.

With that said I still consider Fermis in the current form as failed cards for average Joe.

 

[kylew]

On the other hand (to be a little objective in the ongoing discussion) nVidia may have released a very capable GPU indeed. Let's just assume the prices of 480s drop down to £400 at least and we start seeing these GPUs with watercooled blocks just below £500, we might have the fastest single GPU ever which does not have all the cons of the current referenced versions of 480 - heat output of a volcano, noise of a jet and power draw of... I have to stop it, bills will remain the same or likely be higher when cards get some decent overclocks

I would personally love to see these cards in action and am sure there would be some enthusiasts willing to pay a grand for two of these cards for SLI setup. They would surely do well in DirectX 11 intensive apps in the top high-end systems.

With that said I still consider Fermis in the current form as failed cards for average Joe.

As I said in a previous post:


All it takes is for AMD to release a normal 5870 with a new bios at 1000 core/1300 memory and appropriate volatages, and it's without a doubt the fastest GPU, they don't need to make a new GPU, they can just release new cards with the same hardware and a new bios. This is what makes it the fastest GPU, if the GTX480 was the faster GPU, it would would be able to overclock high enough to top an overclocked 5870.

Why does it take AMD to officially release a card with a modded bios for you to consider it to be the fastest GPU?


According to my above logic, the 5870 is STILL the fastest GPU regardless of what clocks it's currently available in if we assume that it's essentially underclocked as they're pretty much all capable of 1000 on the core.

 

[Mr Krugga]

I'm not arguing with your point of view, just saying that, with some tweaking, nVidia is able to compete with ATI by featuring watercooled graphics.

UK's prices are literally ****ed up, we should see referenced 470s hitting £280 and 480s marking £400 but whatever.

 

[AMG]

I don't think it takes AMD to do that ethoir as in some benchmarks it is ahead by a few percent, all they really have to do is rise those factory clock speeds than sorted.

The GTX480 wouldn't be able to touch a HD5870 with a core of 1Ghz I don't think

watercooling it is just going to cause different problems particularly for those were the chipset, cpu and gpu share the same loop, you would probley need two loops for something like that or a loop with two rads if that is possible