CheckEmoji Community · the emoji forum
🏠 Home 🆕 What's new ❓ Unanswered 🔥 Popular 📡 RSS Members 👥 0 online log in · register
Home › IT › Hardware › Desktops › GPU Buying Guide - (Read first post)

GPU Buying Guide - (Read first post)

Started by Eric Nguyen · · 👁 80 views · 1.5K replies

📡 Subscribe to replies

Participants Eric NguyenAlex Barnes93fadedorca66Jack Lewis36Joseph Young57Alexander Moore67Kyle Jones83Sarah Turner98Joshua Garcia3crimsonstag84Rebecca ParkerRobin Hernandez9Amy Vaughn12Jesse Brown90brightviper13Benjamin King2Larry Ortiz5Terry Scott65shadowjackal16Ryan Gonzalez4Robert Rodriguez2Jeffrey Petersonswiftowl5Megan Brown6 …
Kimberly Gomez34 Kimberly Gomez34 Regular
686 messages
joined Jul 2009
#641 ·
Why wouldn't it be good? 🤷
Based on the data (using an I5 760 in the same rig), you'd be looking at the 7870, gtx 580, or the gtx660Ti running without any bottlenecking. Then you have the 7950 (the i boost version) and the gtx670, which would show a minor bottleneck.
Keep in mind, those numbers below are without any filtering (like AA or AF)—once you toggle those on, the GPU utilization actually climbs even higher. So, both the 7950 and the gtx670 should pass just fine. And all of this is strictly at stock settings, no overclocking involved.

EDIT
Something feels off here—look at the third row: the athlon x4 740, Phenom 960T, and Q9550. According to the table, they can handle anything from a gtx 260 up to a 4890 or 5750. Yet, in the very next row, testing the gts 450, gtx 650, 7750, or 5770 already results in a slight bottleneck.😕

It might be a mistake because that should technically be the "sweet spot" combination—especially since the following row is marked yellow for a mild bottleneck (like the gtx650Ti or 6870), though those should work too. They don't even list a perfect match on this table; it’s almost certainly an error.

Aside from that specific part, everything else seems pretty credible to me.
Larry Ortiz5 Larry Ortiz5 Regular
710 messages
joined May 2020
#642 ·
My bad—I was actually looking at a different chart before and totally panicked for a second. I've got a Sapphire HD 7870 paired with my Proc, and honestly, running everything at stock settings, it performs pretty well enough.
Kimberly Gomez34 Kimberly Gomez34 Regular
686 messages
joined Jul 2009
#643 ·
It’s crucial not to confuse a GPU bottleneck (where the graphics card is the limiting factor) for someone watching these tests—especially when the weakest cards show up in red. For instance, if an 8800GT shows up as red for a certain i5 setup, all that means is the graphics card is being pushed to its absolute limit. If you see the CPU sitting at 40% utilization while the GPU hits 99%, the processor actually has enough headroom to handle much beefier hardware. In this scenario, the GPU is the bottleneck for the CPU, which is exactly what you want to see—you want that graphics card working at full capacity.

Basically, you’re looking to ensure the CPU isn't the bottleneck for the GPU (for example, if the CPU is pinned at 100% while the GPU is struggling at just 70%). You should check this during demanding games for that specific configuration; the more taxing the game, the better the test results will be. Of course, if I throw an old, lightweight game like Commandos 2 at it, it wouldn't be surprising if both components breeze through it with massive FPS while barely using a third of their power.

A lot of this also depends on how well the game is optimized and what’s happening on the screen at the moment (like a loading screen, for instance). To check for bottlenecks, you can use a program like Playclaw3.
crimsonstag84 crimsonstag84 Regular
518 messages
joined Jun 2010
#644 ·
There seem to be a few errors in this table... It shows an Athlon X2 6400+ for the HD 5750 and GTX 260, but then lists an X4 740 for the HD 5770 bottleneck 😂

I suspect the CPUs in rows 3, 4, and 5 might be slightly underestimated. Based on this, it looks like the 3870K actually outperforms the X4 960T 😬
Kimberly Gomez34 Kimberly Gomez34 Regular
686 messages
joined Jul 2009
#645 ·
The "3" is definitely an error. As for the 4 and 5, everything could be shifted right by at least one column to make sense.

And honestly, who knows what hardware they actually used for testing? (For instance, some multiplayer titles just absolutely murder performance) Or what specific criteria they applied? They might consider anything less than a constant 99% to be a bottleneck—even if it just dips to 95% or falls below 90% occasionally. That kind of thing happens even with top-tier processors in something like Battlefield 3 multiplayer. It’s especially suspicious if they didn't test all the CPUs on the exact same games. For example, the 7950 and an i5 were likely pushed hard on the newest, most demanding titles, but I highly doubt they tested the E4400 and a 3870 on something like Far Cry 3 multiplayer. I suspect they might have used games more aligned with that generation's era.

Take this one guy's setup as an example:
An Athlon X2 4400 plus a 3850 DDR3 (both overclocked, which everyone would agree is a standard combo—and it really is)—but in today's CPU-heavy games, the processor is pinned at 100% constantly while the GPU sits at 35%. That is a massive bottleneck. Yet, when another guy uses that same GPU with an E4600 @3GHz, he sees a solid bottleneck in most modern games...
bluejackal21 bluejackal21 Newcomer
1 message
joined Oct 2012
#646 ·
What would be a decent budget GPU? It absolutely has to be low profile to fit inside my HTPC. I just need something capable enough to run FC3.
Benjamin King2 Benjamin King2 Active Member
138 messages
joined Apr 2007
#647 ·
FC3? 😁 There’s no such thing.
bluejackal21 bluejackal21 Newcomer
1 message
joined Oct 2012
#648 ·
So, if I try to cram something bigger into an HTPC case, would it actually work?
image
Benjamin King2 Benjamin King2 Active Member
138 messages
joined Apr 2007
#649 ·
Something like this—but seriously, don't go buying one of those cheap setups with all the messy extension cables attached: http://www.logicsupply.com/products/ga620rse

Personally, I wouldn't touch anything weaker than an HD 7770—and honestly, even that card is going to start sweating once you push past 768p resolution. Look, FC3 isn't some game you can just run on a junker PC.
bluejackal21 bluejackal21 Newcomer
1 message
joined Oct 2012
#650 ·
Is an E6600 running at 3.0 GHz going to cut it?
I noticed the bare minimum for FC3 seems to be around 512MB*
Kimberly Gomez34 Kimberly Gomez34 Regular
686 messages
joined Jul 2009
#651 ·
These are the most powerful low profile options out there.

Even better, if you can track one down somewhere: -
bluejackal21 bluejackal21 Newcomer
1 message
joined Oct 2012
#652 ·
Why are prices dropping while demand is actually picking up?
Benjamin King2 Benjamin King2 Active Member
138 messages
joined Apr 2007
#653 ·
bluejackal21 said:Is a 3.0 GHz E6600 decent enough?

Not really. Just check out this link.
Kimberly Gomez34 Kimberly Gomez34 Regular
686 messages
joined Jul 2009
#654 ·
Look, if you aren't obsessed with every little detail or needing Full HD resolution, just grab that 6670 DDR5 low-profile card and you'll manage to get some gameplay out of it.

Here, check out this video—plenty of these clips featuring the 6670 DDR5 show it running on dual-channel setups very similar to yours.
silentviper56 silentviper56 Member
12 messages
joined Mar 2013
#655 ·
So, I am looking to determine the performance gap regarding
- Sapphire HD 7870, 2GB GDDR5 256bit, 28nm, DVI + HDMI + 2x Mini-D
Sapphire HD 7870, 2GB GDDR5 256bit, 28nm, DVI + HDMI + 2x Mini-DisplayPort, PCI Express 3.0 16x, CrossFire, Core : 1000 MHz, Mem FQ : 4800MHz (1200GHz), DirectX 11, Stream Processors : 1280
how much more powerful it is compared to this
- Sapphire HD 7850 OC, 1GB GDDR5 256bit, 28nm, 900MHz
Sapphire HD 7850 OC, 1GB GDDR5 256bit, 28nm, DVI + HDMI + 2xDP, PCI Express 3.0 16x, CrossFire, Core : 900 MHz, Mem FQ : 4800MHz (1200GHz), DirectX 11, Stream Processors : 1024
as well as the margin between that and this
- MSI GTX660 OC 2048MB GDDR5 , N660-2GD5/OC, 1006MHz Core
MSI GTX660 OC 2048MB GDDR5 , N660-2GD5/OC, 1006MHz Core (Boost Clock:1072MHz), 6008MHz Memory, DVI 2x, HDMI x1, DP x1

However—and this is where I am a bit confused—I have come across several sites claiming that the November 12th driver update, or whatever that specific release was (and I am curious about what exactly this "free download" entails),
significantly improved the 7870, potentially making it substantially better than the GTX660

Furthermore, I would like to know if either the 7870 or the GTX660 will act as a bottleneck for my
- Intel Core i5-3350P 3.1GHz (Max Turbo FQ 3.3 GHz) ,6MB
Kimberly Gomez34 Kimberly Gomez34 Regular
686 messages
joined Jul 2009
#656 ·
silentviper56 said:I'm also wondering if there will be any bottlenecking with the 7870 or GTX 660 when paired with my
- Intel Core i5-3350P 3.1

No, there won't be. As for the performance gap—just remember that you get what you pay for. It's pretty simple as that.
silentviper56 silentviper56 Member
12 messages
joined Mar 2013
#657 ·
So, does that November 12th driver update come as a free download, or is there some kind of catch? 😕
Also—if anyone here actually knows the answer—will a 7870 be able to handle the standalone version of DayZ (which I assume requires
similar specs to Arma III) on high or ultra settings?
Larry Ortiz5 Larry Ortiz5 Regular
710 messages
joined May 2020
#658 ·
The Sapphire HD 7870 is an absolute beast for gaming—you can easily run stuff at 1080p Full HD on max settings without breaking a sweat. I actually use an XFX model myself, so don't worry about your CPU; it’s plenty powerful enough to keep up with the card. As for drivers, well, every manufacturer handles their own software—Nvidia has theirs, and AMD has theirs. Honestly, AMD has really stepped up their game lately, and they're actually outperforming the GTX 660.

If you're looking to grab the drivers, just head over to this site, pick your OS and whether it's a 64-bit system, and you're good to go. http://support.amd.com/us/gpudownload/Pages/index.aspx

Regarding those beta drivers—that basically just means the software isn't quite finished yet and is still in the testing phase. But hey, don't let that scare you; they usually run just as smoothly as the official release versions.
silentviper56 silentviper56 Member
12 messages
joined Mar 2013
#659 ·
Larry Ortiz5 said:The Sapphire HD 7870 is an absolute powerhouse for virtually any game you throw at it—you can easily run things at 1920x1080, or Full HD, with everything cranked up to maximum settings without much trouble. Personally, I use the XFX model myself, and I can assure you that your processor is more than capable of handling it—there won't be any bottlenecking there. Regarding drivers, it is standard practice that every manufacturer provides specific drivers for their own hardware; just as Nvidia maintains their own, AMD does the same for theirs. Furthermore, AMD has significantly improved their driver stability, and in my estimation, they actually outperform the GTX 660.

If you are looking to download the necessary drivers, simply head over to this site, select your specific operating system and bit architecture, and you should be all set. http://support.amd.com/us/gpudownload/Pages/index.aspx

As for those beta drivers, that term essentially implies that the software isn't quite finished—it is effectively in a testing phase—but don't let that worry you too much; they generally perform just as smoothly as the final, stable versions.

Thanks 🙂
A Anonymous Veteran
3.6K messages
joined May 2005
#660 ·
I'm currently rocking an old Nvidia GeForce 8500 512 MB. I can still squeeze out some gameplay from back in 2012, though obviously everything has to be on the lowest settings—some games stutter like crazy, and a few won't even launch, much to my surprise. I'm looking to swap it out for one of these: Gainward Nvidia 2GB DDR3 GT 620, GeForce N610SL 1GB, Sapphire HD 5450 2GB, MSI Nvidia GeForce N620 2GB, or an Asus Nvidia GeForce GT 620. I've got about $75 to work with. It's not a fortune, but I don't need anything super high-end either. If you guys know of any better options, please let me know—I wouldn't mind spending a little extra if it's worth it. Also, I'm currently using an old 17-inch CRT monitor, but I'll be upgrading to a 21-inch LCD soon.

You must log in or register to reply here.

Log in Register

🔗 Similar threads