The 8 core showdown and analysis thread.

[ALXAndy]

uh huh

night...

Trolling my own thread. Yup.

I'll answer the question for you. I decided to keep the Intel. That only leaves me with one AMD cpu but hey, AMDZ FAN80I til I dye mayte.

Seriously,

 

[Mal X]

ha ha, this is gonna end badly

 

[CAT-THE-FIFTH]

Emm,ALXAndy has spent more on his Intel rigs than his AMD rigs,and decides to keep the 8 core Intel one and is being called an AMD fanboi?? Okay???????!!!!!

IIRC,he also has a six core IB-E rig too.

 

[ALXAndy]

Emm,ALXAndy has spent more on his Intel rigs than his AMD rigs,and decides to keep the 8 core Intel one and is being called an AMD fanboi?? Okay???????!!!!!

IIRC,he also has a six core IB-E rig too.

Yup, 3970x that originally replaced the 4.9ghz 8320 that has now been given to my lady and replaced with an 8 core Xeon.

But hey, AMD all the way.

This is what happens when a troll enters a thread and tries to gain some leeway without realising he's savagely missed the point by not reading the actual thread.

Gotta love them skim readers.

 

[McstylisT]

Andy ,

I Don't always agree with your ideology, but have to personally say the information you gathered is interesting. Thank you for posting that.

However, I like to be see the clear story. I think comparing only those two cpu's in those situations only shades one side, because of the lower IPC of the FX and the significant low speed in general of the Xeon. So being brutishly honest it doesn't show the whole story.
To complete this point and clear up the whole situation you should have had a i7 4770/4790K and a stock 4690/4670K in the equation. It would then show the opportunity of the more powerful IPC ( i7 with powerful 8 threads ), but much lower threads in balance (on the i5 with 4x powerful threads) whilst comparing to the well gathered information you currently produced. But hey I understand its not possible and were not made of money, I would just like to see for conclusion.
I generally do favour Intel, I am happy to admit that,, but i more in favour of factual information and the ability to learn. So to be conclusive , as mentioned before im not sure your results give the total scale of lower threaded, but significantly faster IPC balancing and when its available does it prove to be more efficient. ?

But i will conclude if you dont have the funds, and are not in a situation where you can afford and premium processor, then picking one with many threads and as high clock as you can afford is sound advice, but you would guess so.

Thanks for taking the time to do this Andy, and I hope you can honestly understand my points and take them on board.

Regards

 

[Minstadave]

10/10 for effort.

However you've just reviewed an overclocked desktop chip against a stock server chip. This is the daftest comparison I've seen in a while.

You've given us a useful insight into 16 thread gaming. I can't help think the comparison with the AMD chip isn't helpful or surprising.

 

[ALXAndy]

Andy ,

I Don't always agree with your ideology, but have to personally say the information you gathered is interesting. Thank you for posting that.

However, I like to be see the clear story. I think comparing only those two cpu's in those situations only shades one side, because of the lower IPC of the FX and the significant low speed in general of the Xeon. So being brutishly honest it doesn't show the whole story.

Ignore the comparison. Honestly, ignore it. That was just a "Because I could" set of benchmarks and nothing I took overly seriously. If you had a look over my FX 8320 Analysis I did last year you would see that it was far, far more thorough.

I needed to see what the difference was. So hey, brought you guys along for the ride. It's really as simple as that tbh. I mean, who compares a £100 desktop CPU to a £700 server CPU? it's completely discombobulated, but hey, that would be me

LOL looks like I just broke the American dictionary in Chrome with that word

To complete this point and clear up the whole situation you should have had a i7 4770/4790K and a stock 4690/4670K in the equation. It would then show the opportunity of the more powerful IPC ( i7 with powerful 8 threads ), but much lower threads in balance (on the i5 with 4x powerful threads) whilst comparing to the well gathered information you currently produced. But hey I understand its not possible and were not made of money, I would just like to see for conclusion.

If I had an I5, if I had an I7. Sadly I don't and if I build one more rig I'll end up single lol.

When I play games I play games. I don't run FPS counters and I don't bother scrutinising every frame. I'm hyper alert, so if a game is unplayable I'll soon know. But Frame Watching is kinda like bird watching. It can become a bit of an obsession and end up being expensive.

I've long been a bit of a core whore. Mainly because I know how beneficial having say, four pairs of hands would be over just having the one. The more you offload pointless tasks to gaming onto cores (like the fat bloated os sitting underneath them, for example) the more the others can be saturated with what matters.

Servers have used a huge amount of cores for many years now, there's a reason for it. It's far more efficient both in energy terms and productivity.

But hey, if I do ever come across an I5 I'll be sure to thrash it to near death

I will definitely engage in the core reducing thing though. I can also run game benchmarks (more than just Metro) and compare them to the data I still have.

What I wanted this thread to do was to show people just how many cores and threads recent games use. The whole "You only need four cores" argument is beginning to sound a bit old now. I've shown conclusively with every modern game I tested (trust me when I say, I have them all...) that this four cores only nonsense is just that, nonsense.

Off the top of my head? I know for sure that Sleeping Dogs likes a core or 8 too.

I generally do favour Intel, I am happy to admit that,, but i more in favour of factual information and the ability to learn. So to be conclusive , as mentioned before im not sure your results give the total scale of lower threaded, but significantly faster IPC balancing and when its available does it prove to be more efficient. ?

But i will conclude if you dont have the funds, and are not in a situation where you can afford and premium processor, then picking one with many threads and as high clock as you can afford is sound advice, but you would guess so.

Thanks for taking the time to do this Andy, and I hope you can honestly understand my points and take them on board.

Regards

I favour Intel. Every one favours Intel. I've had a metric tonne of Intel CPUs over the years. Sadly since their 'victory' (if you can call it that, but most of the children need that label to attach) their practices have become more and more warped, and certainly not in favour of the consumer.

Absolutely ridiculously priced CPUs at stupid low clock speeds for hundreds of pounds. Why? because they know the children must have that label to carry around.

It seems to me that an awful, awful lot of people in society have to attach themselves to what they see as a better product. The issue is in the world of CPUs the choices are lower than ever. Desktop CPU, Intel or AMD. No Cyrix, Via gave up...

I'll buy what makes financial sense. Just like when I was a kid growing up. My mother did not spoil me (well, she did a little) but we had what we needed. Big old difference between want and need.

Comparing what AMD make to similarly priced products that Intel make makes it awfully confusing. You need to see through an awful lot of BS to make an informed decision. I've been running AMD for about two years now. Was annoyed that the Xeon E3 1220 I bought was locked, and the turbo was locked too. £154, the 8320 obliterated it.

I have been absolutely loathe to spend more than £150 on a CPU for years now. There really isn't any need, not for a gaming PC.

 

[Martini1991]

What good is having lots of cores if they're producing the same/lower results as 4 strong cores? It's having more cores for the sake of having more cores.
So this thread certainly doesn't invalidate the fact an i5 4690K will still power through and smash everything going gaming wise (And doing it better than both the 16 threads of your Xeon, and the 8 threads of your FX83).

Bearing in mind, Xeon is question would set you back 700 quid if you bought at retail.


EDIT :

Put it this way, at 2GHZ, your Ivy 8 core should have the end total performance of a 3770K running at 4GHZ
Ideally, you'd want to test a 4GHZ Ivy against your Ivy 8 core in the games you've put across.

If the figures aren't within margin of error, then the extra core usage is a bit of a fallacy.

That's really what's required.

 

[John24]

Will try and fit it in later mate. Right now the rig is in the hands of its owner

Edit, just realised matey, that's actually a part of the Asus Realbench.. Not sure the scores would be comparable, but it runs H264 encoding as part of the benchmark.

...

There you go. Core use was a bit erratic and it would pause at times for a few seconds. Could be memory speed or something.

Thanks mate, do you still have access to the FX? I'd love to see how that compares.

It's highly likely the realbench x264 test uses a different source and settings so they won't be comparable.

The bench runs the x264 prosess at above normal priority, so that probably explains the GUI being unresponsive, it might affect the task manager graphs aswell. It is possible the thread count needs increasing for all those cores but that's hardcoded so not something you could try.

 

[gpuerrilla]

So a 2Ghz Intel 8 core can compete with a 4.9ghz AMD 8 core...

So the new Intel 8 cores based on Haswell -E that are unlocked, will smash AMD's 8 cores back doors in. That's good to know, guess they will be worth the price tag then, awesome.

Eh.. AMD released it's 8 core cpu's back in October 2011?

Intel's "new" 8 cores are not even out yet, the piledrivers are coming up to two years old. I would expect a 2014 release cpu to beat a two year old cpu. Let's see how much they will retail for!

 

[Troezar]

Interesting comparison. Some frankly stupid replies showing a lack of intelligence but if you can read between the lines there is definitely a good argument for more cores. I'm sure I heard similar arguments when quads first appeared. I'm actually surprised at the performance of the Intel given the low clock speed. I'd be very intersted to see what it would do at 4GHz+ Thanks for going to the effort.

 

[gpuerrilla]

Going brute force on raw clockspeed has had it's day. There will be a sweet spot of many cores for applications that can use it fairly soon, IIRC Sony were on this wavelength with the Cell.

 

[bru]

Going brute force on raw clockspeed has had it's day.

A good while ago, think back to the Pentium 4, that was when Intel gave up on the brute force approach and have been drip feeding us ever since. ( yes I know that is a little unfair as there has been a few good steps, sandy bridge was one.)

 

[JGBR]

I've got 2 Xeon 2.66 4 core, 8 thread CPUS in a old mac pro doing nothing, any good then for gaming on the premise of this post!?

Of interest to some is that ir recently moved from a 980X to a 4790K and I think the 4790 is far better for gaming.

 

[Hagar]

Eh.. AMD released it's 8 core cpu's back in October 2011?

2012, but you are right about almost two years old, mine was one of the first off the van. The price remained remarkably stable for a long time.

 

[ALXAndy]

I've got 2 Xeon 2.66 4 core, 8 thread CPUS in a old mac pro doing nothing, any good then for gaming on the premise of this post!?

Of interest to some is that ir recently moved from a 980X to a 4790K and I think the 4790 is far better for gaming.

I think they're Westmere right?

I'm running a 1.78ghz Westmere with GTX 670 SLI and it does a great job.

I'm quite amazed at how you can physically tell a difference between a 980x and 4790 in gaming..

 

[ALXAndy]

A good while ago, think back to the Pentium 4, that was when Intel gave up on the brute force approach and have been drip feeding us ever since. ( yes I know that is a little unfair as there has been a few good steps, sandy bridge was one.)

For the technology they own they have far, far too many products.

If they hadn't been so mean with Hyperthreading (IE used it on all of their chips) then we would see better support for it. But their rationing creates far too many niche sectors of the market.

I was reading a magazine the other day, and basically it said that back in the day there were two types of CPU. A good one, and a crap one. IE - Intel vs say, Cyrix. There was also a massive difference in price.

Now though? you've got dual core, quad core, hex core, Hyperthreading, eight cores coming, many different types of all of the aforementioned...

It's absolutely ridiculous. Talk about making a market as confusing as possible. No wonder software support is as confused as it is.

Thanks mate, do you still have access to the FX? I'd love to see how that compares.

It's highly likely the realbench x264 test uses a different source and settings so they won't be comparable.

The bench runs the x264 prosess at above normal priority, so that probably explains the GUI being unresponsive, it might affect the task manager graphs aswell. It is possible the thread count needs increasing for all those cores but that's hardcoded so not something you could try.

Sorry it's been removed now. The rig is now in the hands of my lady, hence why I made the decision to stick with the Intel. Far less aggro to run a stock standard CPU that makes the rig quieter.

 

[ALXAndy]

EDIT :

Put it this way, at 2GHZ, your Ivy 8 core should have the end total performance of a 3770K running at 4GHZ
Ideally, you'd want to test a 4GHZ Ivy against your Ivy 8 core in the games you've put across.

If the figures aren't within margin of error, then the extra core usage is a bit of a fallacy.

That's really what's required.

What I did not do was mash together a whole pile of FPS statistics. I could have, but I decided to concentrate more on core use. Remember - I'm paving the way for this 8 core CPU Intel are about to launch by addressing the issues people say they have with massively threaded CPUs.

It's always far more complicated than it actually sounds. Even benchmarking a game you have so many variables that it just makes your head spin. A classic example is Hitman : Absolution. Three times it stuttered at a given point during the benchmark and dropped the min FPS to 23 or so. Then I ran it again, min FPS was 39 or above. This drastically affects the overall 'score' yet in game it doesn't happen.

Was it drivers? just a certain spot during the benchmark? I know that Metro always stutters right around the same frame.

I've told you before, I'm not obsessed with FPS counts. I never, ever study the FPS during a game. I don't have Afterburner installed and I don't furiously obsess over this sort of stuff. I just load up my games, play and enjoy them, turn them off.

I bought Titans and a 3970x for that reason. Chances are I'm not going to get any stuttering or lag now so I can game without thinking about it.

When Crysis 3 launched people were obsessing over benchmarking it. In certain levels AMD 8 core CPUs absolutely wiped the floor with any I5, and even ran the I7s close. Go to a different level all of a sudden the Intel are winning.

To me there are far more variables than what CPU can score the highest FPS.

The bottom line dude? more cores are better. It really is as simple as that.

The new consoles have really slow 8 core CPUs. IIRC it's something like two for the underlying OS, then anything up to 6 for the games.

I know that core handling has changed drastically in Windows 8. So even if that's how it works in Windows 8 then at least you would have six cores not doing two things at once.

It goes back to my analogy about having six pairs of hands rather than just one pair. One pair of hands can take care of Windows, one pair can take care of something else leaving X pairs to do whatever they please. It's just a far, far more efficient way of handling everything, including games.

 

[~>Dg<~]

if you want this bench/test to be excepted go do them both at same clock 2ghz compare the benchies you already done at this speed.

that is all.

 

[Martini1991]

Core usage without figures doesn't actually show anything though.
If the performance is worse, then more cores aren't better.