- Nov 14, 2011
- 10,411
- 5,677
- 136
l2 looks HIGH on this test, a lot higher than what I remember my K7 would achieve.
as for the 3dmark... that's unexpected.. I never really tested HT on 3dmark06, but on Vantage it brings good gains, well at least for nehalem and higher.
this 4600 is a DDR2 or DDR3 card?
anyway, HT seems to work well with some netburst CPUs
http://en.inpai.com.cn/doc/enshowcont.asp?id=7523&pageid=6627
The 4670 (misreported by 3dmark as a 4650) is a DDR3 card- basically the fastest AGP card you can get: http://www.hisdigital.com/un/product2-448.shtml
I noticed that the actual graphics-y 3DMark tests were improved by HT- I suspect that the unusually high load from their CPU tests caused the contention issues Ferzerp mentioned. They throw especially intensive CPU algorithms at every "core" available, so I can see that happening. I'd want to test it under a more realistic gaming benchmark before I turned it off.
Speaking of which- does anyone know any useful game demos which have a benchmark mode I could run? Something a bit more realistic than 3DMark!