Ati's r520 will be 30-40% better than Nvidia's current top (7800gtx)

[fliguy84]

Originally posted by: SumYungGai
The main problem I have is the "ee" sound being added. Why "benchies"? To me, it just sounds retarded and childish. Just like compy or lappy. And "lappy" isn't even much of an abbreviation, and it makes the user look like they're trying to sound cute, or even cool (in their warped sense of it). Abbreviating it is fine, but it doesn't have to sound like a 5 year old made it up.

well, it sounds ok for me. as long as people don't start using the likes of 'bencheese' or 'munchies'

 

[ArchAngel777]

Originally posted by: ronnn

Originally posted by: M0RPH

Originally posted by: ToadkillerDog
Turtle is mocking fanboys who get too serious over hardware. He has created a parody of a fanboy in his posts that many of you are falling for. Some of you are even trying to correct his grammar... He is playing with you guys.

I think you're giving Turtle much more credit than he deserves. IMO, he was not exactly the sharpest tool in the shed. With that said, he was egged on and provoked by certain people here, who then led a campaign to have him banned, which they eventually succeeded at. Just thought you'd like to know that you won't be seeing Turtle here anymore.

Basically agree but think he is very smart, but has some problems with social things. And as you say the environment here is very unforgiving with many attempts to manipulate your emotions. Anyways felt the guy tried hard to be accepted - but no go here.

Ronn, I was starting to think better of you, until this comment. Seriously, please recant... Because, then I won't have to skip over your posts in the future... M0RPH, you as well... Redeem yourselve... Recant and I might be able to think better of you.

 

[nemesismk2]

Originally posted by: Gomce
Apart from the fact that it's a 16pipe card, thanks to the ringbus it will offer better pixel performance, offload the GPU focus on textures pushing resulting in 20% to 40% increase in graphics processing than Nvidia's current offering.

Think of it as X850XT PE VS 6800gt improvement

You'll see!

where is your proof to backup your statement?

 

[xtknight]

Wasn't the r520 supposed to be released today or am I smoking crack? What's the release date for it?

 

[nRollo]

Originally posted by: Amuro

Originally posted by: Rollo

Originally posted by: Amuro
Has these been posted?

http://www.theinq.com/?article=26622
http://www.theinq.com/?article=26617

I don't believe they're true.

Could be true.

The thing about percent increase claims like that is you need to know where the numbers are.

If one card is at 10fps and the other at 15fps, the 15fps is "50% faster" which sounds very impressive but means absolutely nothing.

Now change those numbers to 40fps and 60fps, same "50% faster", but now you have reason to care.

How about 90% faster than the 7800GTX in FEAR but very close in HL2? Is that possible?

Tough to say, I have no R520 to test it.

Like I said, they could have found some setting like 20X15 4X16X where one is 10fps and one is 19fps and say that accurately, but at a setting both can play the actual difference be much less.

Also, I think FEAR is a GITG title, ATI second "big" attempt at vendor relations.

Last, FEAR isn't running "great" on anything as far as we've seen, hard to say what will make it run great. Maybe the R520 has "Shader Instruction XJ9" that only works on FEAR.

 

[Pr0d1gy]

Originally posted by: Rollo
Maybe the R520 has "Shader Instruction XJ9" that only works on FEAR.

LOL!!!! Nice, someone who understands me!

 

[ayabe]

Originally posted by: Rollo

Originally posted by: Amuro

Originally posted by: Rollo

Originally posted by: Amuro
Has these been posted?

http://www.theinq.com/?article=26622
http://www.theinq.com/?article=26617

I don't believe they're true.

Could be true.

The thing about percent increase claims like that is you need to know where the numbers are.

If one card is at 10fps and the other at 15fps, the 15fps is "50% faster" which sounds very impressive but means absolutely nothing.

Now change those numbers to 40fps and 60fps, same "50% faster", but now you have reason to care.

How about 90% faster than the 7800GTX in FEAR but very close in HL2? Is that possible?

Tough to say, I have no R520 to test it.

Like I said, they could have found some setting like 20X15 4X16X where one is 10fps and one is 19fps and say that accurately, but at a setting both can play the actual difference be much less.

Also, I think FEAR is a GITG title, ATI second "big" attempt at vendor relations.

Last, FEAR isn't running "great" on anything as far as we've seen, hard to say what will make it run great. Maybe the R520 has "Shader Instruction XJ9" that only works on FEAR.

Actually Rollo, FEAR is a TWIMTBP title.

Edit: But yes I am also very curious about the 90% faster claim, unless it was an Inq typo and they meant 9%.

 

[Griswold]

I only read the first two and the last page.

I recently got me a 7800GT and I like it so far, but damn the fanbois on page 1&2 make me wish that ATI's 520 or 580 will blow the pants off of the 7800's (but please at a higher price so I dont have to feel bad ) just to shut them up. Sickening.

 

[jiffylube1024]

Originally posted by: Cookie Monster

Rofl.
Back to topic, the core clocked at 625+, i would have to say that the damn card is going to be hot. I wonder if this is going to be another prescott(R520) vs Athlon 64 (G70)

Prescott is hot because of terrible leakage problems that the early transition to 90nm (among other factors) caused. This was compounded by Prescott needing to run in excess of 3 Ghz. Athlon64 is a newer 90nm technology, uses strained silicon (among other new techs) and doesn't have anything like Prescott's leakage problems.

R520 is an even newer technology 90nm part than even A64, so why would you think they would have leakage problems?

I grant you that it's possible leakage (among other factors) were the reason yields were crap in the first tapeouts, but the fact that it's set to release at 625 MHz (and supposedly can go even higher) leads me to believe that heat will be 'under control' . And by under control, I mean it will probably be comparable to the heat output of G70.

 

[sxr7171]

Originally posted by: Turtle 1

Originally posted by: Frackal

Originally posted by: Turtle 1
I begg your pardon my source is just as reliable as sanders .As their the same. Are you insinuating that my source is not reliable. Well I never. My source is infaliable. Your remarks are inferring that I own propriety in Brooklyn and that I am a bridge builder. Nothing could be further from the trueth.Your attempts at putting me in New York are ineffectual and have not caused me to fill my jodhpurs with your BS.

You can't even spell man


What the fvck is infaliable?

spelling mistake infallible (it is incomprehensible to misspell that word I hang my head in shame. I fell like a menhaden out of water. I am opprobrium over my inability of making that perfunctory mistake. But hay it happens

I see you visited a dictionary and pulled some big words out of it. Good for you. You must be really smart. BTW, I'm sure you meant "hey."

 

[DRavisher]

Originally posted by: jiffylube1024
R520 is an even newer technology 90nm part than even A64, so why would you think they would have leakage problems?

I grant you that it's possible leakage (among other factors) were the reason yields were crap in the first tapeouts, but the fact that it's set to release at 625 MHz (and supposedly can go even higher) leads me to believe that heat will be 'under control' . And by under control, I mean it will probably be comparable to the heat output of G70.

But havent both AMD and IBM (basically same tech isn't it?) both been having problems ramping clock without severe heat problems? Opteron 2,8GHz is going to be 120W TDP IIRC, quite a step up from 95W for the 2,6GHz version. And I believe IBM have been experiencing similar problems. I don't know how similar TSMCs tech is to either Intels or AMD/IBMs, but it may well be that the x1800xl will be a nice 'averagely' hot part, while the quite highly clocked XT will be rather hot. Not at all certain of course, but all other transitions to 90nm have shown such problems when ramping clock speeds, so it seems quite possible at least.

 

[RampantAndroid]

Ummm, those inq articles....I don't trust. The grammar in them is so poor.....I don't know why you people must debate this so much....just wait to see the actual 1800XT....then we'll see who wins the videocard war this time....but I must say nVidia has points due to the MSRP of the 1800, as well as ATI being 6 months late.....if ATI cannot match the 7800 GTX 6 months after the fact, they've lost the round altogether....

 

[SuperFubario]

You must be really smart. BTW, I'm sure you meant "hey."

BURN!!!!!!

 

[KeepItRed]

Originally posted by: Rollo
Also, I think FEAR is a GITG title, ATI second "big" attempt at vendor relations.

Last, FEAR isn't running "great" on anything as far as we've seen, hard to say what will make it run great. Maybe the R520 has "Shader Instruction XJ9" that only works on FEAR.

Originally posted by: Rollo
If the performance in these benches holds true when the game is launched in three months, only those with 7800GTXs or high end SLI rigs will be able to play at settings currently taken for granted.

Test it on your 7800GTX SLI system, I'm sure it would run great. The demo ran at 30ish FPS on my 9600PRO with medium settings.

 

[imported_Amuro]

Originally posted by: Turtle 1

Originally posted by: Frackal

Originally posted by: Turtle 1
I begg your pardon my source is just as reliable as sanders .As their the same. Are you insinuating that my source is not reliable. Well I never. My source is infaliable. Your remarks are inferring that I own propriety in Brooklyn and that I am a bridge builder. Nothing could be further from the trueth.Your attempts at putting me in New York are ineffectual and have not caused me to fill my jodhpurs with your BS.

You can't even spell man


What the fvck is infaliable?

spelling mistake infallible (it is incomprehensible to misspell that word I hang my head in shame. I fell like a menhaden out of water. I am opprobrium over my inability of making that perfunctory mistake. But hay it happens

You cannot be "opprobrium" over anything. The word is a noun.

 

[biostud]

Originally posted by: xtknight
Wasn't the r520 supposed to be released today or am I smoking crack? What's the release date for it?

oct. 5th

 

[allies]

Originally posted by: Amuro

Originally posted by: Turtle 1

Originally posted by: Frackal

Originally posted by: Turtle 1
I begg your pardon my source is just as reliable as sanders .As their the same. Are you insinuating that my source is not reliable. Well I never. My source is infaliable. Your remarks are inferring that I own propriety in Brooklyn and that I am a bridge builder. Nothing could be further from the trueth.Your attempts at putting me in New York are ineffectual and have not caused me to fill my jodhpurs with your BS.

You can't even spell man


What the fvck is infaliable?

spelling mistake infallible (it is incomprehensible to misspell that word I hang my head in shame. I fell like a menhaden out of water. I am opprobrium over my inability of making that perfunctory mistake. But hay it happens

You cannot be "opprobrium" over anything. The word is a noun.

haha...

 

[Munky]

Originally posted by: ayabe

Originally posted by: Rollo

Originally posted by: Amuro

Originally posted by: Rollo

Originally posted by: Amuro
Has these been posted?

http://www.theinq.com/?article=26622
http://www.theinq.com/?article=26617

I don't believe they're true.

Could be true.

The thing about percent increase claims like that is you need to know where the numbers are.

If one card is at 10fps and the other at 15fps, the 15fps is "50% faster" which sounds very impressive but means absolutely nothing.

Now change those numbers to 40fps and 60fps, same "50% faster", but now you have reason to care.

How about 90% faster than the 7800GTX in FEAR but very close in HL2? Is that possible?

Tough to say, I have no R520 to test it.

Like I said, they could have found some setting like 20X15 4X16X where one is 10fps and one is 19fps and say that accurately, but at a setting both can play the actual difference be much less.

Also, I think FEAR is a GITG title, ATI second "big" attempt at vendor relations.

Last, FEAR isn't running "great" on anything as far as we've seen, hard to say what will make it run great. Maybe the R520 has "Shader Instruction XJ9" that only works on FEAR.

Actually Rollo, FEAR is a TWIMTBP title.

Edit: But yes I am also very curious about the 90% faster claim, unless it was an Inq typo and they meant 9%.

Why is it so hard to believe? HL2 is not that heavy on shaders, and all modern high end cards breeze through it. FEAR is ruthlessly demanding, and I have a feeling it's because of the longer, more complex shaders, with dynamic branching no less. Since the r520 gpu supposedly uses separate arrays of pixel shaders and texture units, with a scheduler to keep em busy, it could have the potential to run complex shaders much more efficiently, since you dont have stalls within the pipelines when the ALU's are waiting for the texture units or vice versa.

 

[Ronin]

Originally posted by: KeepItRed

Originally posted by: Rollo
Also, I think FEAR is a GITG title, ATI second "big" attempt at vendor relations.

Last, FEAR isn't running "great" on anything as far as we've seen, hard to say what will make it run great. Maybe the R520 has "Shader Instruction XJ9" that only works on FEAR.

Originally posted by: Rollo
If the performance in these benches holds true when the game is launched in three months, only those with 7800GTXs or high end SLI rigs will be able to play at settings currently taken for granted.

Test it on your 7800GTX SLI system, I'm sure it would run great. The demo ran at 30ish FPS on my 9600PRO with medium settings.

Check the FEAR MP Demo benchmark thread to see some of the results (mine included, although I haven't run the new one in SLi yet).

 

[lifeguard1999]

Originally posted by: Gomce
Apart from the fact that it's a 16pipe card, thanks to the ringbus it will offer better pixel performance, offload the GPU focus on textures pushing resulting in 20% to 40% increase in graphics processing than Nvidia's current offering.

Think of it as X850XT PE VS 6800gt improvement

You'll see!

From ExtremeTech: "When we combine all our benchmarks by taking a geometric mean, we get the following performance comparison:

Single X1800 XT..........Vs. X850 XT..........Vs 7800 GTX
1280x1024..................31% faster............3% faster
1280x1024 4x8x..........53% faster..........15% faster
1600x1200..................39% faster..........3% faster
1600x1200 4x8x..........56% faster..........16% faster

Single X1800 XL..........Vs. X850 XT..........Vs. 7800 GT
1280x1024..................6% faster.............6% slower
1280x1024 4x8x..........15% faster...........Even
1600x1200..................8% faster.............7% slower
1600x1200 4x8x..........14% faster............Even

So much for the 20% to 40% faster theory.

 

[ubergeekmeister]

Originally posted by: lifeguard1999

Originally posted by: Gomce
Apart from the fact that it's a 16pipe card, thanks to the ringbus it will offer better pixel performance, offload the GPU focus on textures pushing resulting in 20% to 40% increase in graphics processing than Nvidia's current offering.

Think of it as X850XT PE VS 6800gt improvement

You'll see!

From ExtremeTech: "When we combine all our benchmarks by taking a geometric mean, we get the following performance comparison:

Single X1800 XT..........Vs. X850 XT..........Vs 7800 GTX
1280x1024..................31% faster............3% faster
1280x1024 4x8x..........53% faster..........15% faster
1600x1200..................39% faster..........3% faster
1600x1200 4x8x..........56% faster..........16% faster

Single X1800 XL..........Vs. X850 XT..........Vs. 7800 GT
1280x1024..................6% faster.............6% slower
1280x1024 4x8x..........15% faster...........Even
1600x1200..................8% faster.............7% slower
1600x1200 4x8x..........14% faster............Even

So much for the 20% to 40% faster theory.

Excellent to see some reason and facts in this thread.

Judging from all the reviews I have read, and so far I have read about 8 of them, 20% to 40% better than nvidia and such is just nothing but flame bait and total lunacy.

 

[Duvie]

Exactly vapor!!! All the crap therads we were bombarded with turned out to be dudes...Unless ATi has the best set of drivers in the history of their company and have not released them....

 

[moonboy403]

i'm still waiting for the official 8x.xx driver that suppose to boost the performance of the 7 series by 10% (still praying)
that would almost even things out for the 7800 gtx!!

 

[RampantAndroid]

Originally posted by: moonboy403
i'm still waiting for the official 8x.xx driver that suppose to boost the performance of the 7 series by 10% (still praying)
that would almost even things out for the 7800 gtx!!

That'd be awesome....

 

[Polish3d]

It needs to do it in the AA modes or its virtually pointless IMO.

If it boosts AA performance 5-10%, than we're looking at a GTX being the likely better choice.

If it only boosts non AA performance (which according to the tweaktown review of the 81.82's is the case) than it means jack s*t because no one buys a 500 dollar card to run it w/o AA