ATI, Nvidia Reveal New $250 Graphics Cards

ThinSkin writes "As part of their 'Spring Refresh,' both AMD and Nvidia reveal new $250 graphics cards, the Radeon 4890 and GeForce GTX 275. ExtremeTech takes both cards and runs them through a gamut of gaming and synthetic benchmarks to decide which card triumphs over the other. Long story short, the GeForce takes the cake with impressive performance at its price, while the Radeon didn't show a high improvement over the cheaper Radeon 4870."

Comments

No improvement of the 4870??

[FreakinSyco]

Yet HotHarware tests their 4890 and shows that it outperforms the 4870 in every category...

http://hothardware.com/Articles/ATI-Radeon-HD-4890-RV790-Unveiled/

and I quote:

"In every test, the Radeon HD 4890 (Asus EAH4890) was faster than the 1GB Radeon HD 4870, and the overclocked 4890 (Asus EAH4890 TOP) simply increased the card's overall lead. In comparison to competing offerings from NVIDIA, the Radeon HD 4890 is faster than the GeForce GTX 260 Core 216 overall, but it didn't quite keep pace with the just announced GeForce GTX 275."

Re:No improvement of the 4870??

[Zakabog]

No improvement of the 4870??

and I quote (from the summary):

"Long story short, the GeForce takes the cake with impressive performance at its price, while the Radeon didn't show a high improvement over the cheaper Radeon 4870."

Both articles say the same thing, the GeForce 275 is a better performer, and the 4890 isn't a much higher improvement over the 4870 (though it is still an improvement.)

Re:No improvement of the 4870??

[electrosoccertux]

I didn't read the hothardware article. Did they specify at which resolutions which card wins? Did they test with the newest 185 Nvidia drivers? They're moderately slower than the 182's.

Anandtech, my personal favorite reviewer (none of that 1 paragraph/page + 100 page article nonsense *cough tomshardware cough*) tells a different story.

In case you don't feel like clicking-- 4890 takes the cake hands down on 24" and sub 24" displays (1920x1200 resolution and lower). At 2560x1200, it's a tossup.

Considering you can buy the 4890 right now and the GTX275 won't be available for 2 more weeks, I think it's pretty clear which card to get.

Re:No improvement of the 4870??

[Ecuador]

Replying to my own post above, I actually quickly went through the ExtremeTech article (ok ok take back my /. badge) and confirmed why it is not one of my "preferred" sites. To call the nVidia better they divide the cost by the FPS over the games they tested. Yes, the raw FPS over different games, doesn't need a PhD in statistics to figure out the problem here. Then, they admit that the ATI overclocks very well (with the included utilities from ATI and ASUS), while the nVidia does not, but instead of stating this as an advantage for the ATI card they complain about the cards not shipping at the higher speeds they can easily attain. And finally, they list the nVidia at "$250 street est." price, while the ATI as "MSRP $259" when the latter is right now selling NOW on NewEgg for $230 after MIR or $250 directly. Great Job ExtremeTech!

Beware

[HunterZ]

The Inquirer (I know, they hate nVidia with a passion) is speculating that the GT275 may be a relabeled GT260, except for reviewer cards which may be relabeled GT280's: http://www.theinquirer.net/inquirer/news/599/1051599/nvidia-hoodwinks-reviewers-mythical-gt275s

I guess this is common for ATI/AMD and nVidia to do, but it's the first I've heard of it and it seems awful slimy.

ATI 4890 on linux

[MC68040]

This might be useful to someone like me, Phoronix just reviewed the 4890 on Linux with the ATI catalyst drivers:
http://www.phoronix.com/scan.php?page=article&item=amd_radeon_hd4890&num=1 :)

Confusion

[mac1235]

Hardware site 1 disagrees with hardware site 2! Who can we trust!

Re:Confusion

[644bd346996]

The site that provides more concrete data, particularly about their test methodology. After all, "In god we trust. All others must bring data."

ExtremeTech chose not to test at 2560x1600, despite using a beta driver that significantly alters the resolution scaling performance of the NVidia cards. This means their numbers and thus their conclusions are worthless for anybody who wants to us the higher resolution. On the other hand, they didn't test at 1680x1050 either, so their tests didn't reveal the significant advantage the ATI card has for people with smaller monitors.

HotHardware's review, linked above, did test at both 2560x1600 and 1920x1200, but also neglected 1680x1050. Overall, their testing seemed to be more thorough, but I stopped reading when I noticed that the labels on their first graphs didn't even come close to matching the explanation in the adjacent paragraph. I don't particularly trust numbers when I can't even tell what they're representing.

Anandtech's review seems to test the two new cards in more detail and offers more explanation of how the architectural differences will affect performance. However, they chose to only compare the two cards at that price point, eschewing the context of the more and less expensive products. This makes it harder for the consumer to figure out if it's worth it to spend $250 for a graphics card, but it seems clear to me that Anandtech provides the most reliable data about the two cards, and is certainly more concerned with informing the consumer than deciding for them by proclaiming an all-around winner when there isn't one.

Summary is flamebait

[MooseMuffin]

Nothing in the linked article or the other various reviews of the two cards I've seen today concludes "the Radeon didn't show a high improvement over the cheaper Radeon 4870."

GeForce or Radeon??

[PIBM]

I'm not sure on how to analyze that post after reading TFA. It seems that the radeon beat the nvidia in most of the cases, even at the $ per average FPS..

Thus why is this tagged with nvidia as the winner ?

It's a tie, not a win for nvidia

[Colonel Korn]

Looking at a wider range of reviews, I think we can call this round a draw. That means the real winners are consumers, because the selling point will become price.

Or, if you read the most interesting review of these cards, you'll see why maybe nvidia will skip the price game this time and instead try (and fail) to sell their cards based on physx:

http://www.anandtech.com/video/showdoc.aspx?i=3539

you mean Charlie Demerjian

[MoFoQ]

yea...but don't think it's a "they"...just one crusader (don't think he wears a cape): Charlie Demerjian

He just hates nVidia with a passion.
Supposedly, there was some sort of "tiff" between them...them, like many companies, wanting to limit negative reviews, etc.

either way, grains (not just grain) of salt required for his articles regarding NV.

Re:you mean Charlie Demerjian

[Jah-Wren Ryel]

My impression of charlie's nvidia coverage is that he "sweats the small stuff," "makes mountains out of molehills," etc. But on the big stuff, he's pretty accurate. For example, he had the scoop on the overheating nvidia chips months before it came out anywhere else. That screw-up cost nvidia (and partners like HP) double and maybe triple digit millions of dollars of losses. I would consider the claims in that article to be "big stuff" - sending high-end cards to reviewers and low-end (and mostly unavailable at that) cards to customers is big stuff.

We'll all know for sure in a month or two, but for now, he's got the benefit of my doubt on this claim.

Kind of useless test...

[Anachragnome]

Kind of a useless test from my point of view, that of someone that would be looking to upgrade, for one simple reason.

These comparisons never seem to include the last generation of cards, and thus of no real value to me since I cannot determine how much of an upgrade I would be getting.

I don't care how many fucking cores it has if it doesn't perform better then what I have right now.

Benchmark testing my own machine(as a comparison tool) is sort of useless as well since the REST of my machine may be totally different then what they used.

Which is better then which is an entirely moot point if neither is better then what I have.

Wow

[SnarfQuest]

$250.00. That's more than half of what I paid for my entire system new. The cost of a new system seems to be heavily based on the graphics/monitor. Used to be RAM and disk.

GPUReview

[TypoNAM]

Check out GPUReview's video card compare and see what the theoretical performance differences are:
http://www.gpureview.com/show_cards.php

It does appear that the just announced cards aren't listed on that site yet to compare against unfortunately.

Seems highly unlikely

[Sycraft-fu]

The 275 is something of a "cross" between a 260 and 280. The 280 has 240 shader cores, 1GB of RAM and a 512-bit memory bus. The 260 has 192 or 216 cores depending on the version 896MB of memory and a 448-bit memory bus. They are both 65nm parts. Well the 275 is a 55nm part and nVidia's spec page says it has 240 shader cores, 896MB of memory and a 448-bit memory bus. Hence like I said, a cross between the old two.

Ok well that leave one of two situations for the Inquirer conspiracy theory:

1) nVidia is giving the reviewers cards with more RAM. Possible, but not likley. Also, wouldn't give significantly better results. Turns out that much RAM isn't useful for games these days.

2) nVidia is lying on their product spec page. They are sending 240sp versions to reviewers, and 192 or 216 core versions to the public. Very unlikely, they'd get sued for false advertising.

I just don't buy it. I suppose in theory they could do something like increase the clocks on review cards, but that is real likley to get noticed. Those reviewers know how to run utilities like GPU-Z as well as the rest of us.

I am starting to think nVidia needs to sue Charlie Demerjian for libel. There's not much question his intent is malicious, and he certainly puts out false information. His only defense would be that he didn't know it was false and of course that brings up the question as to why he didn't check, being a journalist and all.

I'll believe this if someone has some kind of real proof, but this seems totally unsubstantiated.

Funny

[Dunbal]

I thought price fixing was illegal in the US. I'm curious as to how the two top competitors manage to release "new" products at the same time and for roughly the same price (within $50 of each other)...

Ohhh wait, must be that "free" market at work again?

Re:Funny

[geekoid]

Both have the same entry price point.
That's not uncommon at all. In fact there is a whole method for calculating it.

There doesn't need to be a conspiracy or collusion or 'price fixing' for 2 similar products to have the same price point.

Re:Funny

[int69h]

Strangely enough McDonald's, Burger King, and Wendy's all sell their sandiwches for roughly the same price as well. Clearly it's the work of a burger cartel.

Re:Funny

[Kjella]

In perfect competition price = marginal cost and unless there's cost differences there shouldn't be a price difference either. Even if you include the write-down costs of assets for a sustainable zero profit operation it still shouldn't differ. The situation is more like this: Around here there's three gas stations very close to each other and they're usually within... in USD, 0.015$/liter. Why? Because it's damn hard to sell overpriced gas if doing a live market survey takes less than a minute. When someone starts with cutthroat pricing they'll all follow and so they all bleed. And none of them can afford to bleed for very long, so when someone raises prices again they follow back up. It doesn't take many rounds of that before you get an implicit collusion - if you don't start we don't start and we all turn a nice profit. Explicit collusion would be all agreeing to add a price hike on top of that. But in all cases the prices are the same across all three, all the way from cutthroat competition to blatant anti-trust violation.

Pay attention

[jgtg32a]

When ATI released the 48x0 cards, their top of the line, they had something like 80% of the performance as the top nvidia cards and cost 1/2 as much.

Prices on cards are dropping.

And as others have stated its a price point.

Re:$200 sound card?

[Sycraft-fu]

Depends on the purpose of the soundcard. In the pro arena, it is number of inputs and outputs, and quality of said inputs and outputs. In the consumer arena it is also I/O quality and to a lesser extent variety, but hardware effects as well.

The effects are pretty simple. Some cards, like the X-Fi, have a DSP onboard and can process 3D spatialization effects in hardware. This is of use for some games. Some games, like say Unreal Tournament 3, only do rudimentary software processing. So if you lack a card that handles OpenAL acceleration, you get only 2 channels and little effects. If you have a hardware OpenAL card, you get full surround and all the effects. Some games can go both ways and simply sound a little better with hardware, or just load the CPU less. Others can't use hardware acceleration at all. It isn't a magic bullet, but useful if you are a gamer (and I am, very much so).

The quality of inputs and outputs is a bigger one, and one you can spend a lot more money on. You are correct that basically every converter these days is 24-bit. However, that isn't the tricky part. The tricky part is all the supporting analog hardware that you use. This is the difference between a system with a high and low SNR and things like that. For example you will find that nothing comes near the theoretical 144dB SNR that 24-bit offers. You may find that a cheap soundcard gets 90-100dB, and it may even by lying about that and reporting the D/A component values, not the final signal, whereas a pro card might get close to 120dB (and really get that).

Also they often have outputs to better deal with special things like low impedance headphones. If I plug my headphones in to the onboard sound, there is audible noise. Why? Well in part the lower SNR but most because its opamp is getting overloaded. The headphones are rather low impedance (40 ohms) and that's more than it can handle. My X-Fi has a headphone port, which has either a buffer to provide more current or a better opamp (or both) so it has no problem powering them and generating no audible noise.

So higher end cards get better overall sound quality, at least if paired with higher quality components later on. Does it matter? Depends on you mostly. Me, I like high quality sound. I have a "home theater" setup for my computer, not computer speakers.

It isn't necessarily something for everyone, but some people like it. Also it may fall by the wayside in consumer systems if HDMI becomes popular. That carries surround audio, in addition to video. So instead of a soundcard you could hook that in a receiver and handle all the D/A conversion there. The digital signal would be output by the video card, of course.