NVIDIA's GeForce GTX 550 Ti: Coming Up Short At $150
by Ryan Smith on March 15, 2011 9:00 AM ESTClosing Thoughts
This is ultimately an underwhelming launch for NVIDIA, but perhaps it’s best we first start with the positives.
The GTS 450 was the first Fermi launch that didn't result in some immediate fanfare for NVIDIA. With performance treading between a Radeon HD 5750 and 5770, the GTS 450 didn’t look good. So if they could be a “most improved” award for a GPU, GF116 and the GTX 550 Ti would most certainly get it. Even though all NVIDIA did was enable a 3rd memory controller and ramp up the clocks, it’s enough to increase performance by 20% - at other segments of the market we regularly settle for less. With these improvements the GTX 550 Ti is finally almost consistently ahead of the Radeon HD 5770.
So what’s the problem? The same problem NVIDIA normally runs in to: pricing. The GTX 550 Ti seems destined to sell based on NVIDIA’s name and market presence more than it will sell based on performance characteristics. Not having a reference card muddles our results some, but ultimately it’s clear that AMD’s pricing has caught NVIDIA flat-footed.
Indeed the GTX 550 Ti is faster than the 5770 - by around 7% - but then the GTX 550 Ti costs 36% more. At the other end of the spectrum is the 6850, which is 7% more expensive on average for 25% better performance. Even the GTX 460 768MB is going to gnaw at NVIDIA here so long as it’s still on the market; it’s 15% faster and yet it’s $20 cheaper. It’s with a dash of Alanis Morissette irony that while having so-so graphical performance the GTX 550 is a remarkable compute card compared to similar AMD cards, but at the same time a CUDA memory bug sliped by before the product shipped.
In these situations NVIDIA reminds me of Intel in the sub-$200 market before Sandy Bridge was released: gross margin first, competition second. AMD is quite willing to cut prices to the bone, NVIDIA is not. As a result on these lower-end products AMD has quite the performance lead for the price. This of course is NVIDIA’s choice, but so long as they choose to go about pricing products this way they’re going to play catch-up to AMD.
In the end the GTX 550 Ti just isn’t a compelling product at $149. At that price you’re much better served by ponying up the extra $10 to pick up a 6850 for much better performance – and if the Zotac GTX 550 Ti AMP is similar to other GTX 550 Ti cards – lower power consumption and less noise. Alternatively the GTX 460 768MB is an absolute steal while it’s still available.
Meanwhile partners like Zotac are left in a rough spot. At $169 $155 the GTX 550 Ti AMP closes the performance gap with the 6850 by some, and at $5 more than a stock clocked GTX 550 Ti is quite a good deal for 10% better performance. But ultimately it's only $5 less than a notably better performing card, the 6850. However the fact that so many partners are doing overclocked cards speaks well of GF116’s overclockability. More significantly it’s quite remarkable that these overclocked GTX 550 Ti’s can get so close to the GTX 460 768MB – a card with a much bigger GPU with many more functional units to work with. With these factory overclocks, the GTX 550 Ti could almost be a decent replacement for the GTX 460 768MB. Pricing is the enemy however – these guys can only lower prices if NVIDIA lets up on the $149 MSRP for the stock clocked GTX 550 Ti.
Finally, we certainly haven’t forgotten about NVIDIA’s interesting memory arrangement with the GTX 550 Ti. It’s a shame that they won’t tell us more about how they’re interleaving memory accesses on this unique design, but hopefully they’ll open up in the future. It’s something we’re definitely going to revisit once the CUDA memory bug is dealt with, and hopefully at that time we’ll be able to learn more about how NVIDIA is accomplishing this. If this is the start of a long term change to memory layout by NVIDIA, then getting to better understand how they’re interleaving memory accesses here will be all the more important to understanding future products.
Facebook
Twitter
RSS
79 Comments
View All Comments
HangFire - Tuesday, March 15, 2011 - link
For a while, AT listened and included the 8800GT with most tests. This was a great baseline as most people understood where their card fell in, compared to the 8800GT.AT has since decided (again) that all of us play nothing but the latest Dx11 games in Dx11 mode with all the goodies turned on, and the only folks upgrading already own Dx11 cards anyway.
mapesdhs - Tuesday, March 15, 2011 - link
Very true!
I've been collating performance results to compare older cards to newer
models as and when I can. Google: "Ian PC Benchmarks", it's the first link
that comes back (Blinkenlights site), then select, "PC Benchmarks, Advice
and Information". Note though that Blinkenlights is a mirror, my main site at
sgidepot is always updated first and more often.
I've included lots of 8800GT, 4890 and GTX 460 1GB data so far and I've
just obtained a 9800GT, 3850 AGP (should be a giggle!) and intend to obtain
various other older cards, including a GTX 275/285. I also have an X1950
Pro AGP (don't giggle, I got better results than reviews of the PCIe version).
Platform wise, I'm trying to put together a socket 775 build and also an
AM2/AM3 setup (I've already obtained a Core2Duo 6850 and Q6600
Core2Quad, though no mbd yet). And I'm adding further P55 examples, eg.
I've obtained an i5 670 and will be including an i5 760 aswell. All this done
on a stupid small budget btw (eBay madness), so if anyone has S775 or
AM2 parts they don't want, feel free to contact me. eBay is not exactly
bargain central anymore. :\ If you're after the highest price though, eBay
is best. Or of course free donations are welcome! 8) (I'll cover the postage;
I'm in the UK) I want to create a spread of data that will be genuinely useful
to people.
I don't have Crysis or Metro33 to test with, but I've used a good range of
freely available tests (recommendations welcome; I'm not going to use AVP
though - I had a look, thought it was rather poor). When I have the time I'll
also add real game tests using my own tests, focusing more on older titles
as that's a common issue people have (I'll be testing with Oblivion, the
1st Stalker game, CoD WAW and a few others).
I'm also including pro apps as & when I can since I do have a number of
borrowed Quadro FX cards to test aswell (580, 1500, 1700, 4500, 5500,
5600, etc.) which will all be for sale once the tests are done. So far I've
done some tests on the 1500 and 5500, but until I sort out a proper certified
X58 setup (for SLI) the results won't be fully fleshed out (Dell T7500
barebones on its way, need parts). Interesting to compare gamer & pro cards.
Note that I'm not massively familiar with older gfx cards, so suggestions are
welcome as to what I should include and/or look for. Feel free to email with
ideas (contact page is on my site, just email my Yahoo acccount). Don't post
here though as that'll only clog up the thread.
Lastly, I'm also putting together a standard X58 setup in a little while, but
first I want to sort out the older systems.
Oh, for those commenting about DX11 on older cards, that's absolutely
true, which is why whenever possible I run each test in all three modes,
ie. DX9, DX10 and DX11.
Ian.
PS. If there happens to be anybody in the Edinburgh area who has a card
they'd be willing to lend me so I can add results for it, please let me know.
You can visit and see for yourself. I'm in the Corstorphine/Clermiston area.
medi01 - Tuesday, March 15, 2011 - link
At least this time it doesn't make you color blind, and bar colors make sense (on most charts) unlike in AMD notebook review.Samus - Tuesday, March 15, 2011 - link
...but so does the GTX460. The 550 comes close to the 'stock' 460 when it is radically overclocked, just as the 460 can beat $200+ cards when it is radically overclocked.I appriciate the overclocking 'potential' and coverage, but ever since the eVGA GTX460 FTW review, AT has been dilluting the true nature of these products with overclocked cards carrying heavier weight in the charts than they should.
Your older reviews (<2009) always had a overclock section, omiting the overclocked nature from the rest of the charts. I liked that.
I just don't like seeing overclocked cards reviewed. They are limited runs and YMMV; the eVGA 460 FTW was available for less than a month after you reviewed it, and has since been replaced twice with the Superclocked, and now the Superclocked Extreme Edition, all of which has had varying GPU/BUS/MEM clocks at prices in excess of $80 over the stock cards. That's BS.
mapesdhs - Tuesday, March 15, 2011 - link
Actually the FTW is still easily available, I bought another two last week for a PC I'm
building for a friend.
Ian.
Ryan Smith - Tuesday, March 15, 2011 - link
A lot has changed since 2009. The biggest of which is that NV and AMD have both given manufacturers more freedom in their designs, and simultaneously manufacturers have been looking to further differentiate their products beyond the cooler and price. Factory overclocks are how they're doing it - it allows them to build a card with a higher performance level for little extra cost, increasing their gross margin while filling small holes in the market.Truth be told it creates a bit of a hassle for us as this results in a different card/clock combo every $10, but clearly it's an effective strategy for the manufacturers. At the same time I get why it frustrates you guys, which is why we don't include these cards on our long run charts. But when it comes to reviewing custom cards it's going to be rare to see cards without a factory overclock - most enthusiast cards now have a factory overclock, and what the manufacturers are willing to sample.
On the plus side, as mapesdhs has already noted, manufacturers are getting better about availability. These cards will never have the kind of long term availability that reference clocked cards do (largely due to the fact that it's a single supplier versus many), but many of them are available through the primary market lifetime of the card (which is to say until it's replaced by a newer GPU).
mapesdhs - Tuesday, March 15, 2011 - link
Has to be said though, I didn't expect the FTW to still be that easily available,
but it is.
However, the earlier poster is also correct that there are slightly lower clocked
alternatives from EVGA that cost less, in one case the core/shader clocks are
the same, just a bit slower RAM (the SSC version). Shop around, and note that
sometimes minor differences in prices can be negated by varying shipping costs
between suppliers. I know one company that keeps offering 'special' deals, but
their shipping costs are so high that they're usually more expensive overall than
alternative sources.
I bought the FTWs because that's what my friend wanted, basically a replica
of the system I built for myself.
Ian.
DrPop - Tuesday, March 15, 2011 - link
I love this site and all the reviews are generally very good.However, I am at a loss as to why this and all other GPU reviewers still use aged compute benchmarks such as folding @ home, etc.
Could you PLEASE start running some BOINC tests with QUALITY, optimized, MODERN code for the latest GPUs, so that the world can see the real "number crunching power" each GPU possesses?
Examples of this would be DNETC on BOINC (highly optimized), or Collatz, etc.
I am quite sure you will be surprised at how the computing bar graph will look - it will be very different than the graphs that come out of your current, aged compute code suite.
Thank you!
Ryan Smith - Tuesday, March 15, 2011 - link
It's true Dnetc is highly optimized (and I use it as a torture test because of that) but it's actually kind of a poor benchmark. It's purely compute bound to the point where cache, memory, etc have no impact. For our compute tests we want benchmarks that stress all aspects of the GPU, so that means it not only needs to be compute intensive, but memory intensive, cache sensitive, etc. Otherwise it's just a proxy for GFLOPs and a best case scenario for the VLIW5 architecture.With that said, I am completely open to suggestions. If you know of other programs that offer a decent benchmark and work on AMD and NVIDIA GPUs, I'd like to hear about it. We'll be refreshing the suite in the next couple of months, so now is the time to make suggestions.
HangFire - Tuesday, March 15, 2011 - link
At least the new 550 is marginally faster and lower power idle than the 450. Someone buying-up from one to the other will get a small boost, and even if power demands are slightly higher, the difference is so small that they shouldn't have to buy a new power supply. If they complain they didn't get a big boost, well, buy something with a larger second digit.After all the naming shenanigans Nvidia has played in the past, they should be commended for (at least) getting the name of the card right.
Memory bandwidth is a very important buying comparison for me. I only buy middle end cards with higher than 50GB/s bandwidth, and high end with more than 100GB/s. This is a form of future proofing. I know I can always turn down detail and still get the frame rates (unless it is a very poorly written game). I would settle for 98GB/s. I would not settle for 32GB/s, or some and some.
Oh, yeah, still no comment from AT on intro-time Linux driver support. Why not at least ask, why give Nvidia shelter on this point?