Nvidia GTS 240 Review (Games & Overclocking) – It Didn’t Survive | Benchy Tests


Hiya everyone! Benchy here from Benchy Tests.
Now, if I’ve managed to get this video out on time, it should be out on Christmas day.
But yeah today, we’re going to be testing this. The GTS 240 from Nvidia, with 1GB of
GDDR3 memory. Now, you might be wondering what even is a GTS 240. You might never have
even heard of it before. And that’s because it was an OEM only card. It only ever came
in packages like a pre-built PC for example, so this never was available for general sale.
With the GTS 240, there were 2 revisions of it. The first revision was built on the 65nm
fabrication process, with this one, the 2nd revision, being built on the 55nm fabrication
process. The differences between those 2 fabrication processes is that the smaller node, this one,
with the 55nm fabrication process, will use less power than the 65nm one. Specs wise they’re
pretty much the same otherwise. As this is an OEM card, you problem don’t really know
what the performance is like. Although, if you’re into your graphics cards you might
already know because it is from the Geforce 200 series of cards. But yeah, with that aise,
let’s get on with the video! The GTS 240 was based on the Tesla 2 architecture,
and the gpu featured 754 million transistors, built on the aforementioned 55nm fabrication
process. It ran at 675MHz on the core, with a 1620MHz shader clock. There’s also 1GB of
GDDR3 VRAM on a 256-bit wide bus running at 1100MHz (or 2200MHz effectively due to the
nature of GDDR memory). The G92B based GPU featured 64KB of L2 cache, and had 14 shading
multiprocessors, totalling 112 shading units, cut down from the GTS 250’s 128. It only needs a single 6 pin power connector,
and will fit into cases with limited space with ease due to its single slot cooler. And
although it technically has directX 11.1 support, it won’t be able to even launch some directX
11games, due to its lack of support for shader model 5. Before the card died in a manner that absolutely,
positively, totally didn’t have anything to do with me, I did manage to get an overclock
and some games tests in, just not as many as normal. And first up is GTA V which was released back
in 2015 on the PC, and remains a great benchmark today. We’re using the lowest settings with
16X anisotropic filtering in both tests. 720p was clearly a lot smoother than 1080p
was able to manage. There was however not much stuttering at all in the city, in fact,
at 1080p there was only one occurence of it and FPS managed to get up to 36 at points,
with drops down to 24fps happening as well. The overclock (//edit: meant to say 720p,
not “the overclock”) showed no stuttering here at all. Both were smooth on the highway towards the
desert area, however the game would dip to 24fps in the desert itself, and at times was
pretty jarring to watch. Again the overclock (//edit: meant to say 720p, not “the overclock”)
was far better here and overall ran great throughout. 1080p showed an average of 31fps, with 24
and 21fps 1% and 0.1% lows respectively, with 720p showing an average of 48fps, a 54.8%
boost in average framerates, and 36, and 33fps for its 1% and 0.1% low framerates respectively.
1080P showed multiple spikes in its frametimes throughout the test of up to 40ms, and a couple
more at 110-120ms as well. 720P had multiple sub 30fps dips throughout and had some 40ms
frametime spikes as well, but these were unnoticeable. In CSGO, with no anti-aliasing, the GTS 240
could max out the settings at 1080p, and ran with no issues at all. There were no lockups
or stutters, and overall the game ran great. It ran so well in fact, that there was no
point in testing resolutions lower than 1080p. The 240 managed average, 1%, and 0.1% lows
of 68, 40, and 35fps respectively. It had a few dips below 30fps according to the frametime
graph, but these weren’t noticeable during the test. As the GTS 240 only has support for up to
shader model 4.0, most newer games aren’t going to run at all. But one that will run
no bother at all, at least in terms of hardware compatibility, is Skyrim. Specifically the
original, and not the 2016 remaster. This is being run using the High graphics preset,
with 16X anisotropic filtering and no anti-aliasing. 720p, as you can see is far smoother than
1080p can manage. There’s occasional but noticeable stutter at points on the run into Riverwood
near the start of the game, which 720p had no issue with other than 1 occurrence where
a dip to 51fps from the games hard 60fps limit was noticeable for a second or 2. 1080p struggled with inconsistent frametimes
in the 1st 1/3 of the test, and had moments in Bleakfalls Temple where the stuttering
was atrocious, which I was down to the game’s enforced use of vsync, which can be disabled
in the games configuration files, but may or may not have unintended effects on gameplay.
Using 720p fixed these problems in the temple which ran pretty smoothly throughout the test
and sat on the game’s 60fps limit for pretty much all of it. 1080p averaged 56fps across the test, and
had 1% and 0.1% lows of 30 and 28fps respectively. 720P averaged 60fps, and the 1% low of 54fps,
shows just how much smoother it was than 1080p. There were spikes in frametimes at 1080p of
around 45-55ms, and 3 more notable ones of 73, 85, and 101ms respectively. 720P had some
40-50ms spikes as well, as well as some around 70-80ms. To improve performance at 1080p you
could try turning down the settings to get rid of the stuttering vsync causes. So now we move on to overclocking the GTS
240. At stock, as I mentioned during the intro, it runs at 675MHz on the core. Using MSI Afterburner,
I maged to push this up to 755MHz before it became unstable. Shader clock wise, I maged
to push it from 1620MHz, all the way up to 1836MHz. Now, this was unstable in CSGO, so
I had to drop it down to 1782MHz to get it stable. Memory wise, I couldn’t get that much
more out of it than the 1100MHz it was already at, but I did manage to boost the clock up
to 1215MHz on the memory. The overclock managed a decent improvement
at 1080p over stock clock performance in GTA V. It still had a couple instances of stutter
(one in the city and one out in the desert area) but could now get to as high as 42fps
throughout the city, with lows up from 24 to the high 20s here. 720P, like stock, performed
great here again, and could now hit as high as 65fps, dipping no lower than 40. The desert area of the game showed similar
increases in performance for 1080p as the city did. FPS could still dip to the high
20s, but was a far less jarring experience than it was at stock clocks. 720P, again much
like stock, was no issue here at all, and wouldn’t drop below 43fps here. Overall, the overclock showed a decent improvement,
although given how well 720p at stock performed, the improvement wasn’t particularly noticeable. 1080p averaged 34fps, a 9.67% increase over
stock’s average at the same resolution, and managed 1% and 0.1% lows of 26, and 24fps
respectively. 720P got its average up from 48, to 53fps, a 10.42% improvement on stock
clocks. 1080P at various points in the tests had spikes in frametimes of up to around 45ms,
some more around the 50-64ms mark, and one of 123ms. 720P only had a few sub 30fps dips
throughout the test which were unnoticeable. As mentioned, the shader clock had to be dropped
down to 1782MHz for CSGO for stability reasons. But given how well stock clocks performed,
taking a little out of the overclock isn’t going to do any harm to performance here.
The game still performs great. It shows no stuttering or locking up whatsoever, but there
was one occasion where it would noticeably drop to 40fps while aiming down the sights
of one of the weapons. We didn’t encounter this at stock as there weren’t times where
we even looked down the sights of the weapons. Fortunately this is the only issue. And as
you can see, the 1% and 0.1% lows saw a small boost, but the biggest jump in numbers came
with the now 77fps average, a 13.23% improvement. Although, average framerates aren’t everything
as you also need to take into account the frametimes and how consistent (or inconsistent)
they are. You could for example be averaging 60fps, which would suggest smooth gameplay,
but if you have frames on screen for 16.7ms (which equals 60fps) and then have a frame
that stays on screen for 33.3ms (which equals 30fps) for example, you might be averaging
60 overall, but you’re going to notice those spikes in frametimes as micro-stuttering,
making a game uncomfortable or even unplayable. With Skyrim, despite having a lot less variance
in its frametimes, and less frequent spikes, 1080p still suffered from noticeable stuttering
on the run into Riverwood. It also had trouble in the Riverwood Trader as well where it was
noticeably stuttery on the upper level of the building. 720P had no issues though and
maintained the game’s 60fps limit throughout the entire test. In fact, the GTS 240 had
so little work to do that the GPU usage was often below 50%. Inside the Temple, it was
much the same for 1080p as it had been at stock, with noticeable stuttering at points
and lows of 34fps here. Again, you could obviously reduce the settings to get a smoother experience,
but it does perform relatively well considering the resolution and settings being used. 1080p managed 57,30, and 29fps for its average,
1%, and 0.1% low framerates respectively. Not much of a gain on stock, but the game’s
hard limit of 60fps kind of makes it near impossible to show much a boost in benchmark
figures. 720P averaged 60fps, the same as stock, and 1% and 0.1% lows of 57, and 30fps
respectively. The 0.1% low for the overclock is actually lower at 720p compared to stock
but performance was good enough for it to be unnoticeable anyway. To conclude, the GTS 240 isn’t particularly
great for 1080p gameplay, unless you’re looking at games such as CSGO. GTA V and Skyrim had
their problems, and given how well 720p performed in both of these games, it’s probably a good
idea to stick to that resolution and up the settings a bit, or even put up the resolution
as well, just maybe not 1080p. Before we finish up today, I’d like to give
shoutouts to Patreon supporters TazFaeDodge and ShadowIntheVoid for helping to make all
of this possible. If you enjoyed this video, please consider leaving a comment and a like,
and subscribing to my channel if you’d like to see more content like this, and maybe even
sharing the video with people you think may enjoy it. And although there’s no obligation
to do so, you can support me in creating these videos through my Patreon at patreon.com/benchtests
or through Ko-Fi at ko-fi.com/benchytests. And if you do decide to become a Patron, you’ll
gain early access to all of my videos for as long as you remain a Patron. But thanks
for watching my video, it means a lot that you would take the time to watch my content.
Hopefully I’ll see you in the next one 🙂

3 thoughts on “Nvidia GTS 240 Review (Games & Overclocking) – It Didn’t Survive | Benchy Tests

  1. Aww you killed it! Lol I must admit, this performed far better than I was expecting it to, At 720 I could happily play those games. Happy Christmas mate!

Leave a Reply

Your email address will not be published. Required fields are marked *