Going from FX8350 to 4770k

[Davedree]

I've posted these gpu usage pics before in a couple of threads, this was fc3
using a 7950 clocked 1050core-1600 memory, usually its 1175-1600. In fc3 with the stock fx8 it really does stutter badly and feels like something is wrong. At 4.8ghz there's still a bit of slowdown but it's much smoother. I've tested a few games since then and the results are interesting. Bf3 multiplayer although shows the fx8 really holding the gpu usage back, (70%) at 3.8ghz and 4.8ghz , the game runs smoothly and without a fps counter you wouldn't know the gpu was being held back. The 2600k even at stock 3.5ghz holds higher min and average fps but gpu usage does drop down to 80-85%, clock the 2600k though and it holds 95-98% gpu usage. If you want to see a bottleneck, then try running truck simulator 2 on any cpu lol.


fx8 3.8ghz fx8 4.8ghz

2600k ht on stock 3.5ghz 2600k ht on 4.8ghz

 

[gpuerrilla]

Much better data. Any more games with that Dave?

 

[Choimaloy75]

That's good, what about Crysis 3?

Trying that,had some problems with crysis 3 with sound issues in xfire think it has to do with v sync.

 

[Choimaloy75]

Sadly, it is what I have always said regardless of martin thinking I am some kind of uneducated fanboy. It has been mentioned in many other threads that it is easy to cherry pick games that benefit both sides of the coin here so choose wisely.

If you can distinctly see a large margin of difference then fantastic for those that it has benefited them. It would also help hugely if you shared the total spend in changing to the new system so there is a fair contrast in how much money was exchanged when 'upgrading'.

As martin has now asked it would be better to now bench some of the more generous multi-core demanding games to get a wider spectrum of performance.

Cost me just under £120 to change,I sold my 8350 and sabretooth for £200 and picked up a returned gigabyte z87 oc for £100 and a 4770k For £218 so I think it's been worth it, just to add done a thief benchmark and got an extra 40fps over the 8350.

 

[Kaapstad]

Cost me just under £120 to change,I sold my 8350 and sabretooth for £200 and picked up a returned gigabyte z87 oc for £100 and a 4770k For £218 so I think it's been worth it, just to add done a thief benchmark and got an extra 40fps over the 8350.

It is good that it has worked out so well.

 

[gpuerrilla]

That's good considering all that GPU horsepower. Did you get any more tests done?

 

[CAT-THE-FIFTH]

That board is being pushed quite a bit though - its a £35 to £40 one with a 4+1 phase VRM.

 

[gpuerrilla]

That board is being pushed quite a bit though - its a £35 to £40 one with a 4+1 phase VRM.

Considering a half decent 990FX board is £75+

 

[Davedree]

That board is being pushed quite a bit though - its a £35 to £40 one with a 4+1 phase VRM.

It's quite a capable board though for a matx though.apm mode, Switchable modules, core unlock.
8pin 12v atx, 4 mosfets per 2 capacitors per each coil, heatsinked. Put it this way it has more potential than the bigger brother stinker known as ga990fxa ud3 rev3, although the rev 4 is good now.
Don"t get me wrong you can't beat an Asus digi6+2 or good 8+2, this board is not high end but even if you push it too hard it has regulation built in, this occurs around 1.45v at 8Cores fully loaded so you can't blow it. Trust me i've given it abuse for nearly 4 months now
It comes down to silicon lottery of course, but 3.8ghz on my fx8320 only needs 1.18v, 4.2ghz needs 1.25v,etc. Except handbrake no application or game is going to max the cores on cpu usage, if it did then it's time for an upgrade.

Which £75 990fx board are you stating just out of curiosity? If msi gda65 then this 78lmt embarrasses it as msi doesn't clock well or offer offset voltage overclocking, the asrock extreme 3 is almost comparable though.

 

[CAT-THE-FIFTH]

The Asus and Gigabyte enthusiast 970 motherboards are well under £70 now.

 

[Davedree]

The Asus and Gigabyte enthusiast 970 motherboards are well under £70 now.

£70 there around yes I agree, but back in time to nov/dec 2013 it was a different story. To fit my needs for my situation I wanted to see how far a £38 MATX board could be pushed and compare phenom II vs fx 43,fx63,fx83 clock for clock and overclocking, What's the problem with that? I'm very skilled and knowledgable in particular amd overclocking, have been using amd since the k6II 500.
iF you've searched previous posts of mine way back you'll see I recommended buying 8+2 or digi for the fx bulldozer/piledriver. People weren't so aware of the throttling issues back in those days, and 4+1 boards at that time usually didn't even have a heatsink on the vrms and barely supported phenom II x4's. The infamous 4+1 situations were the msi's as they didn't have any vrm over-current regulation, and the asrock extreme 3 which didn't have a heatsink or regulation.

Most people still think 4+1/2 boards can't even support an fx8, or that they can't even overclock. I've proved that in the 78lmt's siutation it can offer midrange overclocking potential, providing you don't take the ****
Those that don't believe it'll survive, well we'll see but 4 months in it's still holding 4.8ghz because my cpu happens to be a good one and doesn't suck juice. I've even completed benches of cinebench at 5ghz,
If you need over 1.4v volts and your fx8/6 struggles to scale over 4.6-4.7ghz at this voltage then you need a sabertooth or a later refined fx cpu, end of.

Yes the 970 higher end boards at £70+ are slightly better boards for double the price, The cheaper 970 boards are not any good.
the higher end 970 vs silicon lottery will hit the same 4.6-4.8+ higher on an fx8 comfortably,

I needed a board that was cheap to unlock a Callisto x2 and cope with a moderate fx 6/8. I didn't fancy spending at the time £90 on an asus ma597 evo r2 when it doesn't feature core unlock. If i didn't need to do comparsions between phenom II and fx then I would have gone with an evo R2.
Back in november/dec the prices and choice of boards and stocking availability of the boards were a different situation, to now. It was £130 for an ga990fx ud3 rev4, The inferior ga970 ds3p was £50-£58, the asus m5a97 £60-70 is only 4+2 the £55-60 m5a97 le R2 unheatsinked vrms and no core unlock.

Now though we have the £70 970a ud3 8+2/ or m5a97 evo r2 6+2 £75
These bar the core unlocking support for the asus would be my choice for budget.
But lets not forget the 78lmt is micro atx, and for a little board packs a punch for its price, you cannot dispute that

 

[gpuerrilla]

Cost me just under £120 to change,I sold my 8350 and sabretooth for £200 and picked up a returned gigabyte z87 oc for £100 and a 4770k For £218 so I think it's been worth it, just to add done a thief benchmark and got an extra 40fps over the 8350.

Read this thread which contains:

I have an Intel i7 4770K overclocked to 4.5GHz and Mantle actually has lower average FPS in this game for me.

2560x1600 maximum details apart from SSAA off

DirectX
Min 41
Max 118
Avg 77

Mantle
Min 51
Max 98
Avg 68

So Mantle is ~12% slower on average. The difference is that Mantle gives ~25% higher minimums and ~20% lower highs. This keeps the game looking and feeling much smoother since the frame times remain tighter.

Mantle is showing signs of complimenting the FX. Minimums have improved, which is if you are playing said titles I wouldn't have been hasty to switch.

 

[Avenged7Fold]

Mantle certainly doesn't make any of my rigs lost frames. That is a really weird result.

I get boosts regardless of whether its on my 8350 or 4770k.

 

[Martini1991]

Mantle gives me boosts on games, whether i am on my 8350 or 4770k.

Same (Although given the benchmark is a poor indication of end performance in the game and matters little)

 

[gpuerrilla]

Mantle gives me boosts on games, whether i am on my 8350 or 4770k.

Which is fantastic. The thread is about moving from one to the other though, and as he has listed a mantle supporting title as his latest game, he could have done nothing and played it just fine!

Ed - nice ninja edit A7F!

Same (Although given the benchmark is a poor indication of end performance in the game and matters little)

This just comes across as typical when factual information is now coming through.. can't please some folk!

 

[Avenged7Fold]

Well i figured if i left 'boosts post' in, people wont believe me. I have been open about my AMD and 4770k rigs but loads of people seem to just deny results till their blue in the face regardless of realistic or not.

Ill change it back if you want, its no less true before or after the edit. Its just this thread is a stones throw from becoming another AMD vs intel thread and we have plenty enough of those. Even neutral comments about peoples own set ups is seen as baiting by many forum members.

The amount of cash that went into the Crossfire, a 4770k was a good upgrade. Sure he can achieve decent playability on titles with that set up but the performance he stands to gain on demanding titles is huge! I expect the rig will outlast any game in his current collection, so personally think its money well spent.

If people are scoring lower using mantle on a 4770k, i would put that down to a driver issue they are having, rather than an issue with mantle or the i7.

 

[Martini1991]

If people are scoring lower using mantle on a 4770k, i would put that down to a driver issue they are having, rather than an issue with mantle or the i7.

I saw a few posts about people mentioning that, lower averages, higher mins.

For whatever reason my thief has locked to 60fps, I ain't bothered messing with it yet as I won't start the game yet, I saw gains in minimum and average from mantle to directx, although I'm with you about it being a driver issue, or even the thief patch issue, people forget mantle isn't truly viable yet, on some set ups it doesn't work properly. From what I've used of it, I'm very impressed and haven't had any issues my end.

Either way, the benchmark isn't indicative of end game performance (and Mantle crossfire is broke in thief), actual game frames are quite a fair bit higher.

EDIT : Kazam.

I also see gains across the board with Mantle over Direct X (Again, I still think the benchmark is largely irrelevant to gameplay)

This is my R9 290 - 50mv @ 1050MHZ/1200MHZ (Pretty quiet)

I've taken pictures on my phone so you can/can't see the OSD and also I screenshots don't always work

Here's the DirectX 11. Could have swore I got higher mins before I updated the driver......................... *Tin foil hat*) ;

Here's Mantle

Higher FPS across the board (Derp)

 

[TheyCallMeAndy]

This will be great for you, i doubt you will regret your move at all AMD Chips with high end cards can sometimes bottleneck but the 4770k definitely wont bottleneck you.

 

[Choimaloy75]

I'm over the moon with the move to a 4770k a lot smoother gameplay and higher fps.

 

[Martini1991]

I'm over the moon with the move to a 4770k a lot smoother gameplay and higher fps.

And you're still on the 60HZ monitor that "Would see no gains"?