| Page 1 of 2 |
|
|
 Posted: Fri, 22nd Nov 2013 10:55 Post subject: CPU question |
|
 |
I have a question about this cpu G3420 3.2GHZ/3M/ . Whats the performance in games, is it really bad or mediocre at least? And how is it compared to an ancient e6750duo
|
|
| Back to top |
|
|
|
|
| Posted: Fri, 22nd Nov 2013 11:12 Post subject: |
|
|
it is much faster than Core 2 Duo. In games which arent heavily multithreaded it perform better than AMD FX series for example. But what i have noticed, games are nicely using multiple cores now, so its not really good anymore.
|
|
| Back to top |
|
|
Werelds
Special Little Man
Posts: 15098
Location: 0100111001001100
|
| Posted: Fri, 22nd Nov 2013 11:30 Post subject: |
|
|
| Breezer_ wrote: | | it is much faster than Core 2 Duo. In games which arent heavily multithreaded it perform better than AMD FX series for example. But what i have noticed, games are nicely using multiple cores now, so its not really good anymore. |
What, Breezer 
It's better than an E6750, yes.
But perform better than AMD's FX series? Even the bottom Piledriver, FX-4300 will still easily beat it. |
|
| Back to top |
|
|
Stige
Posts: 3544
Location: Finland
|
|
| Back to top |
|
|
Werelds
Special Little Man
Posts: 15098
Location: 0100111001001100
|
| Posted: Fri, 22nd Nov 2013 11:59 Post subject: |
|
|
Where you getting that from, shit like notebookcheck? 
http://www.cpu-world.com/Compare/237/AMD_FX-Series_FX-4300_vs_Intel_Pentium_Dual-Core_G3420.html
http://www.cpu-world.com/Compare/239/AMD_FX-Series_FX-6300_vs_Intel_Pentium_Dual-Core_G3420.html
That one? Where the 4300 somehow is 10% faster than the 6300 even though the latter has a slightly higher turbo mode?
Those sites are purely theoretical, please do not use them for any sane form of comparison. This thing is not like an i5, it is a much lower power CPU. In extreme cases, like CineBench, it will probably score a few more points, but that's a synthetic benchmark. Even in CineBench, its advantage over a 4300 (or 6300) is probably around 15% at most.
Don't get me wrong, I'm not saying that the FX-6300 is better, because that really depends on the situation, but you can't just flat out claim that in games that are "not heavily multithreaded" it will beat any FX series. In fact, "not heavily multithreaded" translates as "2-3 threads", in which case AMD *will* take the lead because theirs turbo up to 4+. The only point where a CPU like these G series will beat the FX series, is in 100% single threaded situations, such as CineBench.
@ nightfox: if you can somehow get that for free, by all means. But don't waste money on it yourself, it's not going to be worth it, even though it is very cheap. If you do get it though, make sure to get a motherboard that'll let you potentially upgrade to an i5 or i7 down the line and make good use of that.
|
|
| Back to top |
|
|
|
|
|
| Back to top |
|
|
Werelds
Special Little Man
Posts: 15098
Location: 0100111001001100
|
| Posted: Fri, 22nd Nov 2013 12:39 Post subject: |
|
|
Yeah that's perfectly reasonable 
Just don't expect massive jumps. It still is a dual core and for heavier games that still won't be enough. In games where you're CPU limited you will get a nice boost though. Most UE3 games run on 3, CE3 does 4 and Frostbite just eats up all the cores you throw at it almost
That said, I used one of this thing's predecessors in a PC I built for my granddad and it is fantastic value. As a standalone CPU I do think it's better value than the FX series, so don't take my remarks above the wrong way 
|
|
| Back to top |
|
|
|
|
| Posted: Fri, 22nd Nov 2013 13:17 Post subject: |
|
|
Werelds you clearly dont understand the word "when game is not heavily multithreaded". Lets say game what is using only 2 cores, which will win, the FX or the Intel? How hard it is to understand, ofcourse FX will win when game is using more than 2 cores. Single threaded performance is much better on the Intel, no talk can change that. Now since games will probably start using 4 cores at minimum efficiently up to 8 cores, the 2 core intel will be shit tho.
Here you can look some gaming benchmarks: http://pclab.pl/art54829-5.html (these games are cpu heavy and uses 2 cores as its best).
|
|
| Back to top |
|
|
Werelds
Special Little Man
Posts: 15098
Location: 0100111001001100
|
| Posted: Fri, 22nd Nov 2013 14:20 Post subject: |
|
|
"Not heavily multithreaded" in that benchmark means "not multithreaded at all".
FSX, first of all, stems from an era where multi-core support was exceptionally bad and in itself already is horribly optimised on top of that.
The rest are all single threaded games. They do fire off a few threads thanks to DirectX (for things like sound), which thanks to Windows' scheduler spreads out the load somewhat, but if you look at the CPU scaling for each of them you'll see that they only use 1-1.5 core in reality. Each of them scales like crazy with frequency as well. They are all even worse on CPU usage than GTA 4. The core all runs in a single thread, so you end up with a single core maxed out and them some stuff on whatever other cores you may or may not have. Adding more cores will do absolutely nothing for them, but giving them a 500 MHz bump will do wonders.
Take a look at Metro 2033 for example (image below, from that same post). That was built to use two cores and it fully does so (which is what I call "not heavily multithreaded") but going from 2 to 4 will do absolutely jackshit. On top of that, unlike the games you listed, this is not CPU bound.
Don't get me wrong, this Intel CPU will likely outperform the low end AMD CPUs in the majority of games. But your "not heavily multithreaded" argument is just wrong, because what you meant to say is single threaded CPU bound games. The little bit of usage you get on secondary cores are threads are not because of game programming.
The second a game really does use at least two cores, having 2 slightly weaker cores at 4.0+ will edge out the two stronger ones at just 3.2 because the load balances out. You're looking at a total of 1.6 GHz worth of extra ticks then.
|
|
| Back to top |
|
|
Frant
King's Bounty
Posts: 24643
Location: Your Mom
|
|
| Back to top |
|
|
Stige
Posts: 3544
Location: Finland
|
|
| Back to top |
|
|
|
|
|
| Back to top |
|
|
|
|
| Posted: Sun, 1st Dec 2013 12:10 Post subject: |
|
|
Tho in real world you dont notice these "10%" performance bumps.
|
|
| Back to top |
|
|
|
|
| Posted: Sun, 1st Dec 2013 12:37 Post subject: |
|
|
| Breezer_ wrote: | | Tho in real world you dont notice these "10%" performance bumps. |
In gaming, no you wouldn't much. Yet you aren't going to notice much difference between a stock 2600k and one overclocked to 4ghz (Standard overclock that is easily accomplished on just stock cooling) when it comes to gaming as well though.
So either you make that 20% leap from buying the newer product that is using less energy and is on a newer socket standard with more motherboard features (Such as native USB 3.0, PCIE 3.0) or you have a 2600k sandy system using more power to do the same and missing out on those features.
I own both system types. I prefer the Haswell, and as much as I can overclock my 2600k in real world usage my Haswell is just as good but I've got more USB 3.0 ports (Which I actually use) among other things.
Do I wish Haswell overclocked as good as Sandy? Yes. Reality is though that even with my 2600k I didn't need to overclock. I overclocked because I could. Doesn't mean I was really gaining much in real world usage from it though besides better scores in benchmark suites that don't even reflect reality.
|
|
| Back to top |
|
|
Werelds
Special Little Man
Posts: 15098
Location: 0100111001001100
|
| Posted: Sun, 1st Dec 2013 12:55 Post subject: |
|
|
You lot do realise that with Haswell you can do 4.5 just the same right?
It doesn't overclock _worse_ than Ivy, it just doesn't overclock _better_ (which is what everyone was hoping for). And when you put Sandy, Ivy and Haswell all at 4.5 GHz (which are common targets for all three on air), you still get the 10% per generation, so Haswell is ahead of Sandy by 20%.
Does it matter for games? Not really, no.
Mchart is wrong though, because the 1.2/1.1 GHz bump from 3.3/3.4 (for the i5) to 4.5-ish does make a difference. It will not give you better maximum FPS, but it will bump your minimum up in a lot of engines/games. |
|
| Back to top |
|
|
|
|
|
| Back to top |
|
|
Werelds
Special Little Man
Posts: 15098
Location: 0100111001001100
|
| Posted: Sun, 1st Dec 2013 13:10 Post subject: |
|
|
"High enough" - I beg to differ. Plenty of games where the OC makes the difference between 20 or 30 minimum. That's significant to me. Maybe you don't care about that, but I do as I notice every little stutter.
|
|
| Back to top |
|
|
Stige
Posts: 3544
Location: Finland
|
| Posted: Sun, 1st Dec 2013 13:16 Post subject: |
|
|
| Werelds wrote: |
You lot do realise that with Haswell you can do 4.5 just the same right?
It doesn't overclock _worse_ than Ivy, it just doesn't overclock _better_ (which is what everyone was hoping for). And when you put Sandy, Ivy and Haswell all at 4.5 GHz (which are common targets for all three on air), you still get the 10% per generation, so Haswell is ahead of Sandy by 20%.
Does it matter for games? Not really, no.
Mchart is wrong though, because the 1.2/1.1 GHz bump from 3.3/3.4 (for the i5) to 4.5-ish does make a difference. It will not give you better maximum FPS, but it will bump your minimum up in a lot of engines/games. |
You can hit much higher clocks on same cooling on Sandy than you can on anything else if you don't delid.
Now if you delid, you can hit around same clocks on Ivy but Haswell is way worse at overclocking than Sandy or Ivy.
And gaming performance has pretty much no difference between any of them under same speeds, only in benchmarks.
|
|
| Back to top |
|
|
Werelds
Special Little Man
Posts: 15098
Location: 0100111001001100
|
| Posted: Sun, 1st Dec 2013 13:31 Post subject: |
|
|
| Stige wrote: | | You can hit much higher clocks on same cooling on Sandy than you can on anything else if you don't delid. |
Where the fuck do you and other people keep pulling this bullshit from?
http://www.overclock3d.net/reviews/cpu_mainboard/intel_4670k_i5_haswell_review_overclocking/2
http://www.bit-tech.net/hardware/2013/06/12/intel-core-i5-4670k-haswell-cpu-review/7
That's 4.8 and 4.6. I can't be bothered to really go back and find the reviews for Sandy and Ivy, but IIRC Tom got his 2500K to 5 GHz, 4.8 on Ivy. Bit-Tech got their 2500K to 4.9x (they fiddled with the BCLK ), 3570K to 4.8. So this is pretty damn close. Definitely not "much" lower.
The big problem Haswell has is not its overclocking potential. It's the higher variance between samples. There are huge differences; which may mean the difference between 4.6 or 5.0 on air (because there are plenty of people who've got theirs at 5.0 on air). Those differences exist for Sandy and Ivy too, they're just not as big and common as with Haswell. |
|
| Back to top |
|
|
|
|
|
| Back to top |
|
|
Stige
Posts: 3544
Location: Finland
|
| Posted: Sun, 1st Dec 2013 13:42 Post subject: |
|
|
| Werelds wrote: | | Stige wrote: | | You can hit much higher clocks on same cooling on Sandy than you can on anything else if you don't delid. |
Where the fuck do you and other people keep pulling this bullshit from?
http://www.overclock3d.net/reviews/cpu_mainboard/intel_4670k_i5_haswell_review_overclocking/2
http://www.bit-tech.net/hardware/2013/06/12/intel-core-i5-4670k-haswell-cpu-review/7
That's 4.8 and 4.6. I can't be bothered to really go back and find the reviews for Sandy and Ivy, but IIRC Tom got his 2500K to 5 GHz, 4.8 on Ivy. Bit-Tech got their 2500K to 4.9x (they fiddled with the BCLK ), 3570K to 4.8. So this is pretty damn close. Definitely not "much" lower.
The big problem Haswell has is not its overclocking potential. It's the higher variance between samples. There are huge differences; which may mean the difference between 4.6 or 5.0 on air (because there are plenty of people who've got theirs at 5.0 on air). Those differences exist for Sandy and Ivy too, they're just not as big and common as with Haswell. |
The Overclock3D CPU was a much better sample than on average imo.
The second link is worse than average, 4.7GHz with 1.32V and not stable is not very amazing consideirng how big a leap there is in VCore the higher you go with Ivy. I think my 3570K was stable at 1.26V @ 4.6GHz but requires 1.51V for 5GHz.
Also "where we could only fire 1.256V though the CPU to crank the clock speed up to a prime95 stable 4.7GHz before the temperatures got too toasty and the CPU started throttling."
= key bit there, they don't meantion any temps at all, but before I delidded my 3570K, even with my custom water loop I would hit TJMax at just 4.8GHz...
Throttling in Prime with only 1.256V is pretty damn shit.
Getting to 5GHz without delidding is pretty much impossible unless you got a "golden sample" of Ivy/Haswell.
My 2500K ran at 5GHz with a 20€ Thermalright TrueSpirit though, didn't hit TJMax in Prime at 1.52V, never went above 75C in daily use.
Cooling Ivy/Haswell at 1.52V without delidding is impossible 
|
|
| Back to top |
|
|
|
|
| Posted: Sun, 1st Dec 2013 14:43 Post subject: |
|
|
| sabin1981 wrote: | | Werelds wrote: | | "High enough" - I beg to differ. Plenty of games where the OC makes the difference between 20 or 30 minimum. That's significant to me. Maybe you don't care about that, but I do as I notice every little stutter. |
Mchart thinks 30fps is fine for some games, so it's no surprise really. I also vehemently disagree about the "not going to notice much difference" with stock and 4Ghz+ overclock in gaming, that's absurd and I guarantee there's a significant difference.
~note~
This isn't a dig at you Mchart, I WISH I could be fine with it as well. I'd have a lot more fun out of PC gaming, cranking up all the settings and just sitting back to play without fretting over frameloss or stutter  |
I buy hardware that runs said games without having to overclock.
And like I said, I saw no benefit from cranking up my 2600k beyond stock clocks in all the games I play. I might have seen a 5FPS improvement across the board, but if I'm already at 60+ FPS it doesn't fucking matter.
Now, I do overclock my GPU as there are some decent gains there in ANY game - And it's easy. 200mhz more on a GTX680 is a decent performance bump. Bringing a 2600k to 4ghz is such a small performance bump that it's not going to be the difference between playable and not playable.
And really, to keep with objective reality let us take a look -
http://www.anandtech.com/show/7003/the-haswell-review-intel-core-i74770k-i54560k-tested/6
Even with the 20% gain that Haswell brings it's nowhere near 20% faster in games. Even if you overclock that Sandy CPU it's not going to make fuck all worth of a difference in playability.
And yes, 30FPS is fine in games as long as that is the minimum you hit. I could give fuck all about having 60FPS in a game like Assassins Creed or batman. 60FPS is only a requirement when speaking of twitchy FPS games.
|
|
| Back to top |
|
|
|
|
| Posted: Sun, 1st Dec 2013 15:05 Post subject: |
|
|
I'm sorry, but since we're talking "objectively" here, I have to fix your quote and make it correctly subjective;
| Mchart wrote: | | And yes, 30FPS is fine in games for me as long as that is the minimum you hit. I could give fuck all about having 60FPS in a game like Assassins Creed or batman. 60FPS is only a requirement for me when speaking of twitchy FPS games. |
I'm not disparaging that, 30fps is fine (for you) and 60fps is only required in certain games (for you) but as has been quite clearly and repeatedly explained; it's not fine for everyone. 30fps actually feels unpleasant to me now, I see and feel the juddering and frameloss and it's physically unpleasant.
|
|
| Back to top |
|
|
Werelds
Special Little Man
Posts: 15098
Location: 0100111001001100
|
| Posted: Sun, 1st Dec 2013 15:07 Post subject: |
|
|
| Mchart wrote: | | And yes, 30FPS is fine in games as long as that is the minimum you hit. I could give fuck all about having 60FPS in a game like Assassins Creed or batman. 60FPS is only a requirement when speaking of twitchy FPS games. |
No, it isn't. I find 30 FPS very tiring to look at, especially on my PC. And like I said, maybe you don't, but I do notice the irregularities you get with buffered V-Sync. I see the annoying animation judder (and before you or anyone else replies: yes, you do get a form of microstuttering with V-Sync if you don't get 60+ - it's technically impossible *not* to have it, look up how V-Sync works).
For you 30 FPS is fine, but that's not for everyone. And that's also where your aversion from CPU overclocking comes from - you don't care as long as you stay above 30. I on the other hand clearly see a difference between having drops to 20 or to 30 - or even dropping to 45 instead of 55. |
|
| Back to top |
|
|
|
|
| Posted: Sun, 1st Dec 2013 15:07 Post subject: |
|
|
| sabin1981 wrote: | I'm sorry, but since we're talking "objectively" here, I have to fix your quote and make it correctly subjective;
| Mchart wrote: | | And yes, 30FPS is fine in games for me as long as that is the minimum you hit. I could give fuck all about having 60FPS in a game like Assassins Creed or batman. 60FPS is only a requirement for me when speaking of twitchy FPS games. |
I'm not disparaging that, 30fps is fine (for you) and 60fps is only required in certain games (for you) but as has been quite clearly and repeatedly explained; it's not fine for everyone. 30fps actually feels unpleasant to me now, I see and feel the juddering and frameloss and it's physically unpleasant. |
That's because you're hitting well below 30. If 30 is your average then there are going to be issues.
|
|
| Back to top |
|
|
|
|
| Posted: Sun, 1st Dec 2013 15:09 Post subject: |
|
|
What Paul said -- and no, I don't go below 30, why would you even think that? Please stop thinking everything you say is gospel.
|
|
| Back to top |
|
|
|
|
| Posted: Sun, 1st Dec 2013 15:10 Post subject: |
|
|
The entire point of this conversation was that someone on here said someone should buy a Sandy CPU based on the fact that it overclocks well.
That is idiotic given all the reasons that have been mentioned.
|
|
| Back to top |
|
|
Werelds
Special Little Man
Posts: 15098
Location: 0100111001001100
|
| Posted: Sun, 1st Dec 2013 15:14 Post subject: |
|
|
Now that we agree on 
I will be getting Haswell in the next few days too, as I'm not about to buy a platform that is not only old, but has been EOL for 6 months now.
|
|
| Back to top |
|
|
|
|
| Posted: Sun, 1st Dec 2013 15:16 Post subject: |
|
|
Hell, as long as it OCs to/past 4.4Ghz then it's already the same as what I have now \o/ I think the "OC cap" or "doesn't OC as well" crowd is basing their opinion on the 5Ghz barrier, something effortless broken on a lot of Sandy chips but requiring hardware mods like delidding for Ivy/Haswell.
|
|
| Back to top |
|
|
|
|
| Posted: Sun, 1st Dec 2013 15:18 Post subject: |
|
|
Like I said, I'd rather have that 20% performance gain off the bat - Not from overclocking. Further, missing out on have native and hassle free USB 3.0 support should be a massive concern to people. USB 3.0 has matured enough at this point that more stuff besides just hard drives is going to be using it. (Remember, USB 3.0 isn't just about the extra bandwidth, each port provides more power as well)
|
|
| Back to top |
|
|
| Page 1 of 2 |
All times are GMT + 1 Hour |
