Best CPU for a budget gaming PC: Athlon X4 860K vs. Pentium G3258

[Steve]

We received a lot of great feedback after recently updating our PC buying guide and then building and benchmarking the systems. Our most affordable build, the $500 Budget Box, caught the attention of many readers, as it's ideal for everyday computing and will handle some lightweight gaming.

If you're adding a dedicated GPU to handle more than casual gaming, however, it makes sense to drop the AMD A8-7650K -- which we chose for its good balance of CPU and GPU performance -- in favor of the cheaper and faster quad-core 3.7GHz Athlon X4 860K or dual-core 3.2GHz Pentium G3258.

Both are identical price-wise with supporting motherboards starting at around $40 and going up to $70 if you want the latest chipset (A88X vs. H97). Your decision will boil down to which platform offers the best performance where you need it most. That's precisely what we're about to find out as we pair them with the GeForce GTX 960 and Radeon R9 285 in 20 of the most popular AAA titles.

Read the complete review.

 

[noel24]

Well, the article brought Us to the conclusion everybody already knew, even AMD, when they decided to build their Quantum prototype off the 1150 platform. What interest Me is that AMD MOBO doesn't need heatspreaders on power section for 95W+ CPU, while Intel's 88W+ design needs one . On a different note, You could make an effort to photograph the setups You used in an article. Test system specs are Z97 & A88X, photos included are of H97M Pro4 (?) and A68HM-E33. Someone could actually consider a purchase based on article's conclusion and MB visuals and be disapointed after seeing Z97 Anniversary in his shopping cart. One does full OC, the other supports Crossfire?

 

[Guest]

No Far Cry 4? Interesting article but without minimum frame rates it only tells half the story I think. Lets face it, an i5 is the minimum if gamers wanna tackle all past games with no worries and hopefully the majority of games a few years into the future without flushing money down the toilet. Grab a lowend pentium or even a celeron if your on a budget and save even more money and just plan on upgrading down the road if your hard up for money now. Anything AMD is just hot, power hungry, and you're stuck with mediocrity.

 

[Squid Surprise]

Scary... always knew that Intel had AMD beat, but thought that AMD was at least competitive in the budget area... while I'm, thankfully, not a budget buyer, it's kind of sad to see that Intel is the choice no matter what you're doing...

 

[Skidmarksdeluxe]

But Steve, Why didn't you compare these chips to that other low budget chip from Intel... the i7 5960X?

 

[EEatGDL]

To Steve [or anyone who knows first-hand]: let's say not being so tight on budget... which one is a better choice for gaming: Core i3 with stock cooler or Pentium G3258 [OC'd] with water cooling? In Mexico both options cost around the same; but my doubt comes that I've read that some games like Far Cry 4 only run in quad-core CPUs, which the i3 tricks into believing it's a real quad-core.

If the Pentium can be OC'd with stock cooler without being noisy loud nor throttling, I could consider it for some build recommendations on budget.

 

[GhostRyder]

Excellent article @Steve, the budget area is one of those areas people need to see at times especially when paired with appropriate cards to match those areas.

One thing though, the problem with the Pentium now is that since its a dual core some games have issues running on it more so than the 860k (Far Cry 4). Makes the decision sometimes a bit tougher when you have to take that into account.

Personally speaking, if I could get a "good" Pentium chip (Like the one I currently have for testing) then tis a great buy. My only problem with it from the ones I have bought and others I have heard about is that the chip can range from being a spectacular overclocker to barely being able to do any overclocking (One chip I had barely hit 4.2ghz with ridiculous voltages while the one I currently have for myself is at 4.5ghz with lower voltages). The Athlon 860k seems to be a little more consistent from what I have seen personally as far as overclocking plus your not as limited on the games because it has 4 cores. It really is a hard decision overall...

 

[noel24]

To Steve [or anyone who knows first-hand]: let's say not being so tight on budget... which one is a better choice for gaming: Core i3 with stock cooler or Pentium G3258 [OC'd] with water cooling? In Mexico both options cost around the same; but my doubt comes that I've read that some games like Far Cry 4 only run in quad-core CPUs, which the i3 tricks into believing it's a real quad-core.

If the Pentium can be OC'd with stock cooler without being noisy loud nor throttling, I could consider it for some build recommendations on budget.

If you don't need specific IO ports or don't care about number of USB3.0 ports, than cheapest motherboard with i3 and stock cooling will cost You less than G3258, Z97 board and custom cooler to handle OC heat. And CPU intensive GTA5 benchmarks show that cheapest i3 will outperform OCd Pentium. Personally I would go with i3.

 

[Guest]

Hello.Some questions.

1)You are using cheap motherboards, so the overclocking room is small.
2) You are comparing a 860K CPU clocked at 4.20 to a G3258 clocking at 4.40 at the same test. Different freq, so how could you say that this is better or worse clocked lower/higher than the other. The comparison must be at the same freq.
3) You are using 95Watt CPU at a cheap motherboard with absolutely not vrm cooling. This causes throttle. How are you sure that cpu is not throttling during tests?

sorry for my english

 

[Guest]

Um....*record scratching sound* -
How did you guys manage to test a 2-core Intel CPU on Dragon Age Inquisition, a game that does not support Dual-Cores at all?
It says it, right on the box...
Did you actually test Dragon Age Inquisition on the intel rig?

If so, why are you not mentioning the extra steps you have to take to make that game (theoretically) run on that cpu? Don't you think it's important to tell your readers that you have to pull some basic-level hacking tricks in order to get it, (as well as some other games,) to even run on a dual-core system? How about mentioning the trend that more and more games these days require a quad-core or better cpu?

If a person decides to play Dragon Age on a dual-core after modifications: A very, very good air cooler or a water-cooler is required. You have to deal with the extra heat generated by a dual-core under such extreme (and unusual) circumstances. (Where two cores carry the workload designed for four.)

So - why are you neglecting to mention the extra (hacking) steps, the extra price (for the pro cooler,) and the inherent lack of gaming capability that all dual-core systems now possess?
Don't you think that adds value to the AMD X4 - that you *don't* have to hack, just to get it to run some of the most popular games out there?
This review has a curious pro-intel slant to it, and a curious smell.

 

[Tibeardius]

Great review. Lots of good info. I built my first gaming rig out of an old Q6600 and a HD 7750. Worked great for less than $300. Hope this encourages new gamers to build their own pc.

 

[Donny Stanley]

Nice work on the article, great to see so many games tested.

I'm interested why you only included the average FPS which only paints half the performance picture, and not frame time variance in the form of the 1% minimum and 0.1% minimum FPS which gives us a much clearer representation of the CPUs performance, as well as the actual experience a user would get with each.

Telling someone they'd get an average of "80 FPS" on one and "74" on the other but not telling them what the worst FPS is that they can expect 99 percent of the time and 99.9 percent of the time, is a bit disingenuous and makes it seem like you're trying to hide something.

 

[Tibeardius]

Nice work on the article, great to see so many games tested.

I'm interested why you only included the average FPS which only paints half the performance picture, and not frame time variance in the form of the 1% minimum and 0.1% minimum FPS which gives us a much clearer representation of the CPUs performance, as well as the actual experience a user would get with each.

Telling someone they'd get an average of "80 FPS" on one and "74" on the other but not telling them what the worst FPS is that they can expect 99 percent of the time and 99.9 percent of the time, is a bit disingenuous and makes it seem like you're trying to hide something.

Or maybe because this is a huge review. 2 different CPUs at 2 clock speeds each with 2 diffident GPUs benched on 21 games. They've given a lot of solid info. Criticizing them for hiding info is pretty laughable.

I agree those are very important tests but maybe just ask about those benchmarks instead of questioning the integrity of the site that just plopped all these goodies on your lap.

 

[Donny Stanley]

Or maybe because this is a huge review. 2 different CPUs at 2 clock speeds each with 2 diffident GPUs benched on 21 games. They've given a lot of solid info. Criticizing them for hiding info is pretty laughable.

I agree those are very important tests but maybe just ask about those benchmarks instead of questioning the integrity of the site that just plopped all these goodies on your lap.

So what? The size of the review doesn't change anything. If you're using FRAPs to record data then the frame time is already there for you, you just need to add it into your graphs. The problem is, this review means nothing if the average FPS on the Pentium is higher but the frame time variance is so low that it is stutter-y, which is something I've experienced first hand when testing the G3258 in modern, demanding titles such as GTA:V and Witcher 3, etc.

Only listing average FPS in a CPU review is sort of like saying "Car A can do 110 miles per hour, while car B can only do 105" That sounds llike Car A is quite a bit faster as the top speed is higher, but if you fail to mention that the 0-60 is nearly 4 seconds slower on car A then you're not painting the whole performance picture, which is disingenuous.

 

[Tibeardius]

So what? The size of the review doesn't change anything. If you're using FRAPs to record data then the frame time is already there for you, you just need to add it into your graphs. The problem is, this review means nothing if the average FPS on the Pentium is higher but the frame time variance is so low that it is stutter-y, which is something I've experienced first hand when testing the G3258 in modern, demanding titles such as GTA:V and Witcher 3, etc.

Only listing average FPS in a CPU review is sort of like saying "Car A can do 110 miles per hour, while car B can only do 105" That sounds llike Car A is quite a bit faster as the top speed is higher, but if you fail to mention that the 0-60 is nearly 4 seconds slower on car A then you're not painting the whole performance picture, which is disingenuous.

Yeah, I didn't need that spelled out. My point was that you just assumed that they are intentionally misleading their readers by not posting every detail instead of giving them the benefit of the doubt. You know what they say about assuming.

 

[Guest]

Yeah, I didn't need that spelled out. My point was that you just assumed that they are intentionally misleading their readers by not posting every detail instead of giving them the benefit of the doubt. You know what they say about assuming.

You're right, perhaps I was a bit too presumptious in my assement. As a hardware reviewer I've become a bit jaded as there is so much bias in the industry and I see articles with vital data left out it just immediately smells bad to me, especially when I've seen very different results in my own testing.

 

[Tibeardius]

You're right, perhaps I was a bit too presumptious in my assement. As a hardware reviewer I've become a bit jaded as there is so much bias in the industry and I see articles with vital data left out it just immediately smells bad to me, especially when I've seen very different results in my own testing.

It's understandable. I've seen it too. Ideally the writer finds the information and updates the article.

 

[Evernessince]

So there's really only two situations you would want an AMD CPU.

1) Ultra budget integrated GPU only
2) Low Budget Machine with high thread requirements.

Otherwise, Intel has had a monopoly on everything else for a long time. Chances are you'll rarely see an AMD cpu fill the two needs I delineated above either, as Intel controls the OEMs.

 

[EEatGDL]

If you don't need specific IO ports or don't care about number of USB3.0 ports, than cheapest motherboard with i3 and stock cooling will cost You less than G3258, Z97 board and custom cooler to handle OC heat. And CPU intensive GTA5 benchmarks show that cheapest i3 will outperform OCd Pentium. Personally I would go with i3.

Thanks for answering. That's exactly the criteria I used for a system I built for a cousin at the end of the last year; I revisited my old doubt with this article coming out. I haven't analyzed too many games comparing both scenarios, but I can say my cousin was more than satisfied running GTA V on his PC -something I'm not sure could have happened if gone the other way.

 

[Steve]

Hello.Some questions.

1)You are using cheap motherboards, so the overclocking room is small.
2) You are comparing a 860K CPU clocked at 4.20 to a G3258 clocking at 4.40 at the same test. Different freq, so how could you say that this is better or worse clocked lower/higher than the other. The comparison must be at the same freq.
3) You are using 95Watt CPU at a cheap motherboard with absolutely not vrm cooling. This causes throttle. How are you sure that cpu is not throttling during tests?

sorry for my english

1. Overclocking headroom seemed good to us and why would you pair $70 CPUs with $200 motherboards?
2. We aren’t, both processors were clocked at 4.4GHz when overclocked. This was listed on the graphs and in the system specs.
3. No the CPU wasn’t throttling when overclocked.

Um....*record scratching sound* -
How did you guys manage to test a 2-core Intel CPU on Dragon Age Inquisition, a game that does not support Dual-Cores at all?
It says it, right on the box...
Did you actually test Dragon Age Inquisition on the intel rig?

If so, why are you not mentioning the extra steps you have to take to make that game (theoretically) run on that cpu? Don't you think it's important to tell your readers that you have to pull some basic-level hacking tricks in order to get it, (as well as some other games,) to even run on a dual-core system? How about mentioning the trend that more and more games these days require a quad-core or better cpu?

If a person decides to play Dragon Age on a dual-core after modifications: A very, very good air cooler or a water-cooler is required. You have to deal with the extra heat generated by a dual-core under such extreme (and unusual) circumstances. (Where two cores carry the workload designed for four.)

So - why are you neglecting to mention the extra (hacking) steps, the extra price (for the pro cooler,) and the inherent lack of gaming capability that all dual-core systems now possess?
Don't you think that adds value to the AMD X4 - that you *don't* have to hack, just to get it to run some of the most popular games out there?
This review has a curious pro-intel slant to it, and a curious smell.

The developer fell back on this decision upon release. It was clear the game could run on Intel dual-cores so they enabled support. This happened a long time ago, there is no need for any hacks which is obviously why it wasn’t mentioned.

Nice work on the article, great to see so many games tested.

I'm interested why you only included the average FPS which only paints half the performance picture, and not frame time variance in the form of the 1% minimum and 0.1% minimum FPS which gives us a much clearer representation of the CPUs performance, as well as the actual experience a user would get with each.

Telling someone they'd get an average of "80 FPS" on one and "74" on the other but not telling them what the worst FPS is that they can expect 99 percent of the time and 99.9 percent of the time, is a bit disingenuous and makes it seem like you're trying to hide something.

Not sure what we would be trying to hide here. The minimum frame rates told the same story.

Yeah, I didn't need that spelled out. My point was that you just assumed that they are intentionally misleading their readers by not posting every detail instead of giving them the benefit of the doubt. You know what they say about assuming.

You're right, perhaps I was a bit too presumptious in my assement. As a hardware reviewer I've become a bit jaded as there is so much bias in the industry and I see articles with vital data left out it just immediately smells bad to me, especially when I've seen very different results in my own testing.

If you were really a hardware reviewer you would know there isn’t ‘so much bias in the industry’ as most sources come up with similar results.

 

[Donny Stanley]

Not sure what we would be trying to hide here. The minimum frame rates told the same story.

There's a difference between the absolute minimum (or maximum) FPS, and the frame times (read: 99th and 99.9th percentile) not only that but if either of those told the same story, why not post them for us all to see rather than have us take your word for it?

If you were really a hardware reviewer you would know there isn’t ‘so much bias in the industry’ as most sources come up with similar results..

The only time results like this are shown is when A ) You don't show all the data and give the reader a clear picture. OR B ) You test in titles that use 1-2 threads, which the majority of modern, demanding titles, do not.

 

[Steve]

There's a difference between the absolute minimum (or maximum) FPS, and the frame times (read: 99th and 99.9th percentile) not only that but if either of those told the same story, why not post them for us all to see rather than have us take your word for it?

Because it’s a heap of data that not everyone wants to sift through to get the same story. If you have evidence that the AMD CPU is better when looking at this data feel free to provide it.

The only time results like this are shown is when A ) You don't show all the data and give the reader a clear picture. OR B ) You test in titles that use 1-2 threads, which the majority of modern, demanding titles, do not.

Interesting conclusion, it’s not like we always include all those metrics and only removed them for this review. I don’t even know how to address the second part of your comment … “we only test titles with 1-2 threads”? Okay well if you say so.

 

[Donny Stanley]

Interesting conclusion, it’s not like we always include all those metrics and only removed them for this review. I don’t even know how to address the second part of your comment … “we only test titles with 1-2 threads”? Okay well if you say so.

I never said the titles you tested here only used 1-2 threads, I was speaking generally about other reports that have shown similar outcomes. Now you're just taking my statements out of context, sort of like the data displayed in this report.

 

[Steve]

I never said the titles you tested here only used 1-2 threads, I was speaking generally about other reports that have shown similar outcomes. Now you're just taking my statements out of context, sort of like the data displayed in this report.

Ohh right that makes more sense, sorry I didn't follow very well. Anyway I suggest if you feel I deliberately or otherwise gave misleading results you shouldn't read my articles. I put a heap of work into this, tested a ton of games and I was generally interested to find the outcome.

 

[Donny Stanley]

Ohh right that makes more sense, sorry I didn't follow very well. Anyway I suggest if you feel I deliberately or otherwise gave misleading results you shouldn't read my articles. I put a heap of work into this, tested a ton of games and I was generally interested to find the outcome.

I can totally appreciate the amount of work you put into this article, testing 21 games with this many different configurations is no easy task and I'm sure it took a considerable amount of time. That being said, while it is totally possible, and even likely that your intentions were not malicious in writing this report, it is still hard to come to the conclusion that you did without seeing all the data. That is all I'm saying.