The NVIDIA GeForce GTX 980 Review: Maxwell Mark 2
by Ryan Smith on September 18, 2014 10:30 PM ESTSynthetics
As always we’ll also take a quick look at synthetic performance. These tests mainly serve as a canary for finding important architectural changes, and with the exception of pixel throughput we are not expecting any major changes for GTX 980 and GM204.
GM204 is designed to have an ever-so-slightly higher triangle throughput rate than GK110 – 16 tris/clock versus 15 tris/clock, and sure enough the GTX 980 comes out on top in TessMark, slightly edging out the GTX 780 Ti. The difference is only very slight here, and though GM204 should be a bit more powerful than GK110 in practice it’s a dead heat.
Moving on, we have our 3DMark Vantage texture and pixel fillrate tests, which present our cards with massive amounts of texturing and color blending work. These aren’t results we suggest comparing across different vendors, but they’re good for tracking improvements and changes within a single product family.
Beginning with Maxwell NVIDIA reduced their texture-to-compute ratio from 12:1 to 16:1. As a result of this change Maxwell GPUs have fewer texture units than comparable Kepler GPUs. Compounding this effect is the fact that Maxwell CUDA cores are more efficient than Kepler CUDA cores, leading to NVIDIA placing fewer cores overall and further reducing the texture fill rate.
As a result the GTX 980 is not texture fillrate competitive with any of the GK110 cards. It is competitive with the GK104 cards, but only because these cards had the same number of texture units at 128. NVIDIA has told us that they believe this new ratio is a better fit for modern workloads, and judging from the performance we’re seeing elsewhere it would appear that NVIDIA is right.
On the other hand, thanks to NVIDIA’s newer 3rd generation delta color compression technology, our 3DMark pixel fillrate performance is through the roof. GTX 980 comes very close to doubling the throughput of our GK110 cards and more than doubles the throughput of the GK104 cards, reflecting the fact that it has 64 ROPs and more importantly has the available bandwidth to put them to good use.
This benchmark in a nutshell is why NVIDIA can deliver chart-topping performance despite having only 2/3rds the memory bandwidth of GTX 780 Ti. By improving their color compression to this point, NVIDIA can significantly reduce their memory bandwidth requirements Maxwell 2, allowing them to do more with less. In real games the result won’t be anywhere near this remarkable since this is a pure pixel fillrate test, but it goes to show that NVIDIA has been able to expand their effective memory bandwidth in concert with their ROP and shader performance improvements.
Facebook
Twitter
RSS
274 Comments
View All Comments
Laststop311 - Saturday, September 20, 2014 - link
I'm going to wait for the custom gtx 980's. It was already throttling from reaching the 80C limit on most games. Blower design wouldn't of throttled if they left the vapor chamber in but they didnt. My case has plenty of airflow so i don't require a blower design. MSI twin frozr V open air design will cool the gpu much better and stop it from throttling during gaming. People rushing to buy the reference design are missing out on 100's of mhz due to thermal throttle.chizow - Saturday, September 20, 2014 - link
Yep the open-faced custom coolers are definitely better at OC'ing, especially in single-GPU configs, but the problems I have with them are:1) they tend to have cheaper build quality than the ref, especially the NVTTM cooler which is just classy stuff. The custom coolers replace this with lots and lots of plastic, visible heatpipes, cheapo looking fans. If I wanted an Arctic Accelero on my GPUs I would just buy one.
2) they usually take longer to come to market. Frequently +3-6 weeks lead time. I know its not a super long time in the grand scheme of things, but I'd rather upgrade sooner.
3) The blowers tend to do better in SLI over longer periods of time, and also don't impact your CPU temps/OC as much. I have a ton of airflow too (HAF-X) but I still prefer most of the heat being expelled from the start, and not through my H100i rad.
4) Frankly I'm not too worried about squeezing the last 100-150MHz out of these chips. There was a time I might have been, but I tend to stick it to a safe OC about 100-150MHz below what most people are getting and then call it a day without having to do a dozen 3DMark loops to verify stability.
Laststop311 - Sunday, September 21, 2014 - link
Did you see the benchmarks. Some games were running in the 900's some in the 1000's some in 1100's. Stuck at these frequencies because the card was riding the 80C limit. As the review mentioned these aren't the same titan coolers as they removed the vapor chamber and replaced it with regular heatpipes. Getting a custom cooled card isnt about squeezing the last 100-150 from an OC its about squeezing an extra 400-600 mhz from an OC as many reviewers have gotten the gtx 980 to OC to 1500mhz. We are talking a massive performance increase from getting the proper cooling bigger than even the r9 290x going from reference to custom and that was pretty big itself.Laststop311 - Sunday, September 21, 2014 - link
Even to get the card to reliably run at stock settings during intense gaming you need a custyom cooled card. The reference cooled card can't even reliably hit its stock clock under intense gaming because the blower cooler without vapor chamber sucks.chizow - Sunday, September 21, 2014 - link
No, you can adjust the Nvidia fan and GPU temp settings to get sustained Boosts. There is a trade-off in terms of fan noise and/or operating temps, but it is easy to get close to the results of the custom coolers at the expense of fan noise. I personally set my fan curve differently because I think Nvidia's 80C target temp profile is a little bit too passive in how quickly it ramps up fanspeeds. I don't expect to have any problems at all maintaining rated Boost speed, and if I want to overclock, I fully understand the sacrifice will be more fan noise over the custom coolers, but the rest of the negatives regarding custom coolers makes the reference cooler more appealing to me.venk90 - Thursday, September 18, 2014 - link
The GTX 980 page on NVIDIA website seems to indicate HDMI 1.4 as it says 3840*2160 at 30 Hz over HDMI (it is mentioned as a foot note). Are you sure about it being HDMI 2.0 ?Ryan Smith - Thursday, September 18, 2014 - link
Yes. I've confirmed it in writing and in person.vegitto4 - Thursday, September 18, 2014 - link
Hi Ryan, great review! There will be the usual HTPC perspective? For example, did they fix the 23.976 refresh rate as Haswell does? I think it's important to know how these work as htpc cards. RegardsRyan Smith - Thursday, September 18, 2014 - link
For this article there will not. These cards aren't your traditional HTPC cards. However we can possibly look into it for next week's follow-up.chizow - Friday, September 19, 2014 - link
I think the definition of HTPC is beginning to change though, and while these may not yet fit into traditional HTPC (Brix and NUC seem to be filling this niche more), they are definitely right in the SteamBox/BattleBox category.Honestly, SteamBox was the first thing that came to mind when I saw that 165W TDP on the GTX 980, we will be seeing a lot of GM204 variants in the upcoming years in SFF, LAN, SteamBox and gaming laptop form factors that is for sure.