Game Benchmark Results
All benchmarks were run with Windows 8.1 64-bit as the OS, and either AMD Catalyst 14.4 or NVIDIA GeForce 337.50 drivers (which were the most current when the testing was performed).
BioShock Ultimate
First off, every combination here managed to average over 30 FPS, so at “low” detail settings (which still look pretty good at 1080p) the game was playable. But look at the bottom four results! Regardless of CPU the R7 250 was completely outclassed by the other cards, which isn’t surprising given this lineup. I just wasn’t expecting it to be this big of a gap. Buy hey, a $79 card can game at 1080p (at least with BioShock, anyway).
Moving up to 60 FPS the R7 260X / Athlon 5350 build produced nearly that average across the test runs, but the same CPU with the GTX 750 managed a slightly higher average at 62.4 FPS, but did have a slight drop in the minimum frame rate. The higher-powered CPUs certainly made a difference with this game, as switching to an Athlon X4 760K or Pentium G3220 brought the average FPS up considerably with these GPUs and would have permitted playable results at higher detail settings.
Things were close at the top, but the overall leader here was the GTX 750 Ti / Pentium G3220 build with 109.4 FPS. (Not bad for a $399 system!)
Metro Last Light
Even at the lowest settings with the official benchmark tool, Metro Last Light proved a lot more demanding for these hardware configurations, and it didn’t provide a very consistent overall experience due to some stuttering on many of the test runs (the minimum observed FPS number is based on the lowest frame rate from any of the runs, and is not an average).
This game appears to be much more CPU dependent, and here we see results from the each GPU suffer when paired with either the Athlon 5350 and the Athlon X2 340X. The R7 250 is still at the bottom, but this time the frame rates just aren’t playable from any of the lower-end CPU combos – unless you consider 24 FPS playable, and to be honest it’s not bad. (Hey, it’s worked for the movie industry for decades after all!)
Essentially for a smooth gameplay experience at 1080p you’ll want to put a little more money into the CPU for Metro Last Light. The results for the Pentium G3220 and Athlon X4 760K were all very playable (other than those random hiccups resulting in the low min FPS numbers).
Thief (DX11)
I won’t get too much into Thief here since I already talked about the lack of Mantle results here. Simply put, if you have the compatible hardware and drivers, Thief is much better on Mantle. In this case, I compared only DX11 results with all tested components for an apples-to-apples comparison.
Thief is pretty tough on lower-end hardware with DX11 (and looks very CPU bound), and the least expensive options at 30 FPS+ (actually 40+ in this case) came from the Pentium G3220 system with either a GTX750 or R7 260X, but the R7 260X is the clear price/performance winner of the group here with an impressive 46.1 FPS, second from the top!
Next we'll look at the synthetic benchmarks.
![[H] takes another shot at the Falcon](https://pcper.com/wp-content/uploads/2019/03/pcper-news.jpg)
“Though the GTX 750 was
“Though the GTX 750 was nearly identical across the game benchmarks, the 260X put up a heck of a showing in the synthetic benchmarks often matching or even beating an overclocked GTX 750 Ti at times (which made me kind of question the game benchmarks I chose). Plus, the R7 260X supports Mantle. You know, that graphics API I didn’t use here?”
So you pick the 260x because it does better in a synthetic benchmark (not something you can f’ing PLAY) and then see Mantle as a bonus even though you did ZERO testing with it?
The GeForce stuff has GeForce Experience (with ShadowPlay – dramatically better image quality than RAPTR), TWITCH streaming, GameStream to Shield, G-SYNC, PhysX, CUDA, better drivers, etc.
Weak PCPer…weak.
The 260X performed
The 260X performed identically in most of the tests to the 750 TI and it costs less, so shouldn’t it naturally be the best choice?
Really? it was ~10-15% slower
Really? it was ~10-15% slower how is that identical? On top of that, 260x also uses 120watts vs only 60 for 750ti. 750 non-TI card is in same price/performance as 260x, on other note with that 750 is a 55watt TDP part.
the 260x is a lot cheaper,
the 260x is a lot cheaper, raptr and obs now have VCE support, and people are saying it looks better than shadowplay,
Shield is not free (for most people), PhysX is a rarity, and do you really think a 750 is fast enough for 1080P rendering + physx acceleration?
Gsync? you mean the overpriced and unavailable thing? people buying budget PCs are not going to buy gsync monitors, oh and the 260x is going to support adaptive sync,
Cuda is for developers, and the Radeon have better OpenCL support I think.
there is no big advantage for nvidia, and Mantle is real, and works nicely with slow CPUs to improve gaming performance.
750 Ti is way to overpriced, a nice card in terms of power efficiency, but performance per $ is more important here.
I don’t know why people think
I don’t know why people think the 750/750 Ti are the greatest cards since sliced bread. They are awesome when it comes to power consumption and heat, but not that great in performance to price. I’d take a GTX 660 over a 750 Ti any day of the week for only a couple teners more. The performance gap between the two is pretty significant.
What are the complainers
What are the complainers going to do when they have to see the real world. I swear, if you don’t like it just say something constructive instead of throwing insults. It really is counter productive.
Totes agree, the fanboyism is
Totes agree, the fanboyism is really getting old! Just buy what works for YOU, and don’t complain when someone recommends something different.
r7 265 150$ on newegg 256bit
r7 265 150$ on newegg 256bit memory
7850 169$ 256bit memory
hd both of them better choices than the 750ti
It’s the better gpu at its
It’s the better gpu at its price point but when building a system you have to consider the savings you can make on the power supply.
Nvidia competes strongly for the budget minded without offering unacceptable performance.
What I really want to see is
What I really want to see is how this stacks up with the Linux version of Metro Last Light and maybe some other games too. Because if you are really budget who is paying for an OS.
Linux is not viable for
Linux is not viable for gaming yet…
people saving money normally pirate windows (oh and small tablets get windows 8 for free), but obviously it’s wrong.
so there is not way to avoid paying for the win license if you are gaming, because linux gaming is to limited, some old games, a few indies!? metro, an extremely bad port of witcher 2, and what else?
There’s this indie developer
There’s this indie developer called Valve, you probably haven’t heard of them, but I think they have a few games that run on linux now.
they are included in the “old
they are included in the “old games” category for the most part.
Do the PS4 and XB1 operate at
Do the PS4 and XB1 operate at low settings or medium? It would have been nice to see how these systems performed at the settings that the consoles use.
BF4 runs close to “high” on
BF4 runs close to “high” on the PS4, but at 1600×900
xbone looks the same but 1280×720…
the 760K + 260x with mantle would probably look a lot like the ps4 version,
You should use G3258 instead
You should use G3258 instead G3220.
it’s useless if the
it’s useless if the motherboard does not support overclock, and overclock ready motherboards are normally more expensive, in that case i3 + $50 motherboard is better for gaming (most of the time) than 3258 + $90 motherboard.
Asus has recently released an
Asus has recently released an announcement informing us that their latest H97, H87, B85 and H81 Series motherboards has now the ability to overclock 4th Generation Intel Haswell Core K Series (unlock) processors, including the Intel Pentium G3258 Anniversary Edition processor.
Others will probably follow.
last year asrock and others
last year asrock and others announced b85/h87/h81 overclock (for K CPUs) motherboards, a few weeks later Intel announced a microcode update to lock it again, so Intel changed their mind about requiring Z chipset for OC?
Don’t be an idiot. All
Don’t be an idiot. All Asrock’s B85, H87 and H97 (not sure of H81 though) can still overclock any K-series processor as of today since last year. If you have no idea of what you’re talking about, please shut up. There’s no such microcode update released by Intel whatsoever. I’ve seen even ridiculous comments on linustechtips that speak of Intel releasing update through Windows Update that disable non-Z OC on motherboard. What a bunch of idiots!
Yes i agree
you are also an
Yes i agree
you are also an idiot
The 760k would do better if
The 760k would do better if overclocked, it is unlocked afterall.
So who is the r7 250 for
So who is the r7 250 for anyways? Is it atleast better than intel hd 4000? Who is the market, what can it do, why is it a thing?
it certainly is, it even
it certainly is, it even beats the 7850K IGP easily.
it can play most current games with lower settings (include resolution) OK, it’s a huge upgrade over the Intel HD Graphics from the Pentiums…
Yep, the R7 250 beats the
Yep, the R7 250 beats the 7850K even if you overclock the IGP core in the APU significantly (I tested the 7850K with the IGPU core at 900MHz). The 250 certainly would be a great option for 1366×768 (or possibly a little higher) gaming for only $80.
Just not able to handle 1080p very well – but 1920×1080 is a million more pixels to pump out vs 1366×768 after all…
I know this is a budget build
I know this is a budget build and all but all I could think of when seeing those horrible numbers for 1080p and low settings was “But for only $100 more, you could have a MUCH better gaming experience!”
Or a week trolling
Or a week trolling Craigslist. A couple lucky trades and I ended up with a 290X for something like $140 in the end!
This round-up is useless
This round-up is useless without the 265…
The r7 265 is a great card
The r7 265 is a great card yes, better than the 750ti, but it’s just a little too pricy to be considered a budget card, especially considering that you can normally find a r9 270 for a few bucks cheaper on sale
It’s not possible to include
It’s not possible to include a lot of valid options when operating within a budget, as I was. The R7 260X is around $30 cheaper than a 265, and I bought every component in this review myself. $150 was the hard cutoff for a GPU, and I originally didn’t want to go that high. The original group was APU graphics, R7 240, 250, 260X plus a GTX 750. I had to improvise and added a 750 Ti as numbers for the bottom cards (APU, R7 240) were unacceptable but I still wanted at least four cards. In the interest of balance I added a second NVIDIA GPU to have two of each. This was back in April before a couple of lower cost NV cards were announced. That’s all.
After looking over TDP’s to,
After looking over TDP’s to, athlon x4 is 100watt tdp, intel g3220 is 55watts, and on gpu side, 260x is 115watts and 750/750ti is 55/60watts, Seems like g3220 and either 750/750ti would be cheaper option not just initial cost but over say 1 year the lower usage of electric.
Yeah, I’m gonna call BS on
Yeah, I’m gonna call BS on that.
http://www.hardocp.com/article/2014/06/23/asus_r7_260x_directcu_ii_oc_video_card_review/9
System power draw 308W with the 260X, 317W with the 750ti.
um 750ti without pci power
um 750ti without pci power plug pull 200 watts? where is the real BS ?
You serious? Nvidia quotes
You serious? Nvidia quotes averages. Reviews that dive into power comsumption have 750 Tis spiking up to 141 watts with no aux plugs. That’s double of what it should be drawing from PCIe spec.
How do you measure that
How do you measure that reliably without worrying it might be another component? Genuine question. If the GPU utilization spikes, the CPU utilization will probably increase, too, possibly other board parts. Are people somehow measuring at the PCIE slot?
My guess is they simply
My guess is they simply measure by total system power consumption at the plug. Not perfect but close enough.
My guess is they simply
My guess is they simply measure by total system power consumption at the plug. Not perfect but close enough.
By measuring through a PCIe
By measuring through a PCIe adapter card.
That eliminates any whole system measuring. They aren’t using your $20-$40 wall plug measurement. They use oscilloscopes with adapters to get the best reading possible for what ever piece of hardware they are testing.
There are multiple problems
There are multiple problems here. The review at hardocp is using a wall plug and measuring total power draw, but they are using a ridiculously overpowered and inefficient power supply for this purpose, a 1200W Antec PSU.
They state there was a 95W idle draw with no GPU installed. That’s a huge overhead. Those high GPU wattage numbers do not stand up with such a setup.
Also, they are using an overclocked version of the 260X vs a stock GTX 750 Ti. Not a fair comparison.
From the linked article on hardocp:
“We tested the power utilization at the wall of the entire system without a video card, and with each video card at idle and full load. For full load power and temperature testing we used real gaming, in this case every game we tested. The power supply used in testing is an ANTEC 1200W High Current Pro. Our system is very lean with only one optical drive and one SSD being powered. Total system wattage at idle without video card is 95W.”
That’s the same way Ryan test
That’s the same way Ryan test power consumption.
Right. I measure system power
Right. I measure system power like that, too. My point is just that when you measure a low-power system from the wall for total power draw, a 1200W power supply like the Antec HardOCP used is a horrible choice.
This was an awesome article!
This was an awesome article! Looking at the latest and greatest is fun, but real world testing, with real world budgets makes me come back to PCPer time and again.
That new pentium anniversary
That new pentium anniversary K edition is a pretty hardcore piece of silicon. Catching up to i7`s in a lot of games after some overclocking and costs 70$.
Sapphire R9 270 is only $160
Sapphire R9 270 is only $160 on Newegg and also Amazon.
I’m really glad you did an
I’m really glad you did an article like this, as this is probably what a large number of home builders are doing. The only think I would say that may help the article is to note that the system prices don’t include optical drive, case, screen, or keyboard and mouse. I end up helping people with budget builds fairly often, and find that they are usually starting with a new case, and want the optical drive to use old games that they have. (LAN parties with old versions of battlefield or Age of Empires II)
I should also note that the
I should also note that the people are usually doing this for the first time, or have a build old enough that they can’t reuse the optical drive.
(No subject)
You make a really good point,
You make a really good point, and it's true that for many the optical drive is still necessary. It's a great thing to have even if it's an external USB drive (what I use). I didn't include the optical drive to keep costs down, and I didn't use one at all for my builds or testing (installed Windows 8.1 from a USB drive). I mentioned that prices didn't include case/operating system, but of course a totally new build becomes much more expensive once you add a monitor and peripherals.
I have this dream of a $300-ish system that plays current games with SteamOS or a Linux distro as a free OS, and it's sad that very often a better deal on a total system could come from Dell or another OEM vendor. At least there are great low-power options with GPUs that don't require an upgraded PSU to make a vendor PC more feasable on the extreme budget…
Why no R7 265 in the testing?
Why no R7 265 in the testing? It is the same price point at the 750Ti?
It was the timing of the
It was the timing of the purchases, not any intention or oversight.
The cards were purchased in April when the R7 265 was more expensive. At current prices it would certainly be the 750 Ti competitior, and of course now the Ti can be found for $139 and there are new GTX 730/740 budget cards as well.
When I ordered these cards the R7 260X was still $149 for most versions, the XFX model was the cheapest online at $129 at the time. 265’s were around $169.
It’s nice to know that I made
It’s nice to know that I made a good choice when I got a new computer with the Pentium G3220 as the CPU.
Of course for the sort of gaming I do, the Pentium alone with the integrated (Iris lite) graphics mostly works fine for me, with my 1280 monitor and AthlonXP-era game collection.
Only Sacred (which doesn’t like Intel HD graphics drivers)doesn’t run right. Everything else runs faster than my old AthlonXP and 7600GS.
I take this back, Duke Nukem 3D High Resolution Pack balks at the opengl check.
Edit: Duke3D HRP works just
Edit: Duke3D HRP works just fine, if you use -forcegl in the commandline. The eduke engine folks were just lazy in their version check…
cool article. but even a old
cool article. but even a old core duo runs those games, i would really like to see this new doggy games like watch dogs , wolfestein etc, running at lower resolution 720p @900p + high settings… do they reach 30 fps?
Can’t you please let me know
Can’t you please let me know what’s the best PC to play games on 180p or more higher .I don’t know how to built one .thank you a lot