1055t vs. i5-750

Hi guys,

My current system is a socket 775 core 2 quad Q9300 running at 3.2ghz,ASUS P5N-d motherboard, 4gb ddr2, H50 cooler, 2x 250gb hdd in RAID 0, and 2x GTX275's in sli. I use this system mostly for gaming (70%), 10% for work ie spreadsheets, text etc, 15% video editing (youtube), 5% web.

The system is new and runs good but I am disappointed in some of my benchmarks, at least where sli is concerned. In 3dmark vantage I get 21116 when running SLI. I get 20834 in non-sli. Those numbers aren't too bad imo for older tech like the 775 cpu's but I obviously have a pretty significant bottleneck going on.

I want to upgrade the mobo/cpu/and ram to reduce or eliminate the cpu bottleneck. Since the gpu's are new SLI is a must. I have $450 to invest for those three components and I plan on using the core 2 quad to rebuild an Antec 900 I have and hope to sell it off for enough to pay off the new parts (plus maybe a little extra), I have a 8800gtx laying around and everything else to set it up, but I digress...

I am torn between the 2 following setups:

ASUS M4N98TD EVO
AMD 1055t 2.8ghz hexcore
4gb gskill ram
$422 shipped

and

ASUS 55i Sabertooth or ASUS P7P55d Pro
Intel i5-750 2.66 quad
4gb gskill ram
$436 shipped

The system has to last for 2-3 years, please take that into consideration. In the future I also plan to upgrade to dx11 cards, possibly the new "gts 450 top" by ASUS.

Thoughts and suggestions? Thanks

I would go with the 1055. Given a good mobo 4ghz should be possible (same for the 750), and 50% more cores means much more power for multicore stuff.

If you want to go sli, you pretty much have to go with Intel. With an AMD cpu the only chipset I would use would be from AMD.

Ooo i missed the SLI, good catch.

For SLI on AMD you pretty much have to use cracked drivers, nvidias AMD chipsets aren't especially good.

The i5 750 would be much faster, especially if you compare max overclock performance between the two chips. Then there is SLI compatability as well...

Epsilon84 said:

The i5 750 would be much faster, especially if you compare max overclock performance between the two chips. Then there is SLI compatability as well...

Click to expand...

xokeman said:

Click to expand...

Lol. Nice post. Very constructive.

i5 750 @ 4GHz would destroy an Phenom X6 @ 4GHz in games. No current game can take advantage of more than 4 cores, so the extra cores on X6 go wasted.

Heck, even at stock speeds (2.66GHz) its about as fast as a 4GHz Phenom
http://www.techspot.com/review/305-starcraft2-performance/page13.html


http://www.techspot.com/review/320-civilization-v-performance/page12.html



Owned.

Those are almost all two or four core AMD cpus there. No six cores listed anywhere.
Care to offer proof on the no games using > 4 cores?

Also note BFBC2 can eat any cores you can throw at it...

An i5 is going to get prison sexed by a well overclocked AMD x6.

Bobnova said:

Those are almost all two or four core AMD cpus there. No six cores listed anywhere.
Care to offer proof on the no games using > 4 cores?

Click to expand...

I take it you're talking about Starcraft II? Because Civilization 5 has a 3GHz 1075T in there, and it gets 28fps compared to 36fps on a 2.66GHz i5 750.

With Starcraft II, I suggest you use your powers of deduction and logic. Look at the difference between Phenom X2 and X4 at 3.8GHz. NONE. Therefore 6 cores will make no difference. More proof that SC2 doesn't take advantage of more than 2 cores here.

As for proof in regards to games and core count, sure. Bit-tech has an article looking at how core count affects performance in games. In appears that in *most* cases, 3 cores are the 'sweetspot', and 2 cores are enough for 'playable' framerates. Quad or hex cores offer little noticeable difference compared to 3 cores.

EarthDog said:

Also note BFBC2 can eat any cores you can throw at it...

Click to expand...

Incorrect. 6 cores show no tangible gains compared to 4 cores.

http://www.bit-tech.net/hardware/cpus/2010/07/05/how-many-cpu-cores-do-games-need/5

Oh...Edit...nevermind, lol

You are totally right.

Its really weird how my frame rates plummeted when I wither disable HT, or when I switched to a dual with HT...

From the article...

In the bottom graph the engine appears highly threaded, with no core left untouched and between 15/20 to 60 per cent usage.

Click to expand...

so my statement:

Also note BFBC2 can eat any cores you can throw at it...

Click to expand...

in response to:

No current game can take advantage of more than 4 cores

Click to expand...

...is correct.

EarthDog said:

Its really weird how my frame rates plummeted when I wither disable HT, or when I switched to a dual with HT...

Click to expand...

Would you be able to take a screenshot of the Task Manager on your i7 with HT on and off? I doubt you will find that 4 cores can be maxed out at 100%, let alone 4C/8T.

From reading gaming forums, I know that Core i3 suffers big framerate dips in heavy firefights in BC2, but speaking from personal experience my i5 @ 4GHz with a HD4890 it runs beautifully (in DX10 mode obviously).

EDited above btw...

i7 860 is gone to another land... literally. so Im on the i3 550 now at 4Ghz and noticing the slowdowns you mention vs 4 cores vs 4cores + 4 threads.

EarthDog said:

Its really weird how my frame rates plummeted when I wither disable HT, or when I switched to a dual with HT...

From the article...

so my statement:

in response to:

...is correct.

Click to expand...

So it spreads the load over the 6 cores instead of 4 cores, with the result being that each core is utilised to a lesser degree. Sure, no core is idle, but it doesn't actually increase the actual framerate as the benchmarks show.

Right. But it uses them. They do not sit idle, contrary to the statement I quoted (at least thats how I thought you meant it). Other games do not do that (see your own link).

Oh well, spinning wheels here.

Ahh, guess we just had different approaches to 'using all cores'. I gather you meant this chart showing CPU usage?

The busier cores are loaded at ~50% on average, and the least busy core is only loaded at ~25%. When averaged out over all the cores, there is only about 40% core utilisation across the board.

Yep that chart.

Absolutely on the different approach. I took it at face value, more or less out of context than within the scope of the post and thread..