Kyle at Hard OCP Looking like an Ass!

[Technoid]

Originally posted by Ant
Sorry folks but I've just had an email about this thread from Kyle so I'm going to have to ban you all and delete it.

[ALBPM]

This is all too funny especially since Kyle is a total AMD suck up.

I thought it was strange that he would make an Intel product look so good.


I always thought Kyle had one of Jerry Sanders hemoroids stuck between his teeth.

[isochar]

The article was written by Steve Lynch. Kyle probably just glanced over it and gave it the ok.

[Wombat]

It wasn't just a 3/8 mistype. Someone specifically said over at xtreme that it went from 178xx to 173yy

[The PIT]

The graph I saw it went from 17829 down to 17329. Of course there may have been more editing going on since.
Anyway they're still acting like sour pusses over there with an IQ of one. Good job Ant doesn't behave like this or there wouldn't be anyone left to post here. Ant Ant why can't I post anymore.
By the way I entered Mac address's in a database and I can tell you the numbers can get completly messed up. The most common is S for 5. Sometimes the address given is nothing like the one that the network card has got.
Oh and I ain't standing up for Kylie.

[SitFlyer]

Originally posted by RichL
I'm just waiting for this story to get on the likes of The Register or The Inquirer. 'Tis daily food to them.

Originally posted by RichL
http://www.theinquirer.net/?article=6290

Didnt take them long

Boy that was quick


edit: This news won't appear in The Register. A quote from Drew Cullen: "Our policy is to ignore publishing spats, of which there appear to be far too many in hardware review land."

[UtwigMU]

What bothers me is their pools:

Intel's Hyper-Threading
- This could be something BIG
- No matter for real computing
- It is a minus

From what I read on Ace's hyperthreading brings 5-10% boost when multitasking or using hyperthreaded apps and 0-5% when using singlethreaded apps.

IMO it's thoughtful way of using the core better = beneficial.

I could not categorize it as the next big thing, a minus (it obviously isn't one) and unsignificant.

And this is not the first dead pool end I saw there.


Also this: let it speak for itself:

The GeForce FX supports AGP 3.0 specification for AGP 8X bandwidth of 2.1GB/sec. In current games you probably won't find much use for AGP 8X over AGP 4X. Perhaps Unreal Tournament 2003 and future games such as Unreal 2 and Doom 3 will benefit from that. With the increasing texture size that games are beginning to use, AGP 8X will be a needed technology. 128MB of local video memory on a video card may not be enough for future games to enjoy playing them at high resolutions with features like anti-aliasing turned on. When the game runs out of local storage, it will need to go into system RAM, and the only way for it to do this fast enough that you won’t notice any hitching in the games is to have a fast AGP transfer speed. AGP 4X has a clock rate of 266MHz and 1.1GB/sec bandwidth. AGP 8X has a clock rate of 533MHz and 2.1GB/sec of bandwidth. AGP 8X is simply an evolutionary step that may be needed in games to experience the smoothest gameplay.

I will indeed not notice any hitching when going from 16-20GB/s internal to 2.1GB/s AGP

IMO Tom is better.

[Technoid]

The AGP spec was somewhat disappointing when it came and has always been a bottleneck since the FSB raised above 66MHz

It’s basically a slightly mutated PCI slot

[thop]

actually HT is pretty neat. the recent c't review explains very good what HT is. what is probably more important than raw figures (they measured up to 50% performance increase in some apps) is the subjective speed feeling. they say that HT greatly improves the responsiveness of XP.

[Wombat]

Technoid, you really underestimate the AGP slot. The direct memory access makes all the difference in the world. It's also not the AGP slot that is the holdback as much as your system's RAM.

HT does have drawbacks. Improper switching will slow down your program execution. Also, just putting it in to the silicon constrains the maximum speed of the execution unit.

[thop]

they weren't able to find anything that was slower with HT than without, although they say that sometimes this can be the case, as even intel admits so. one problem they found are many "cache aliasing conflicts" because of the shared cache, so they say, which can be circumvented with clever programming (_alloca they say, but dont ask me why ) though.

the only thing were HT had no benefit was some heavily optimized SSE testsuite that uses 4 single-precision floatingpoint per instruction (whatever that means). also they didn't use the windows task manager to mesaure the cpu usage but some self coded tool called PerfWatch that reads some internal perfomance counters.

i don't know much about processor design but it is a very interesting read

[Wombat]

No, they're definitely spinning it. Intel's had code run slower with HT, and I know it perfectly well. The "it could happen, but we haven't seen it" thing is bull.

[UtwigMU]

For HT article that is explaining, based on knowledge and sees past benchmarks (can explain why some of them are the way they are.) read aceshardware.

Compare articles from aces, toms, [H] and anand and you'll see for yourself.

[thop]

the c't article is much more in depth, they dont have many benchmarks though (mostly synthetic ones to see what HT can do under optimal conditions). really recommended if you can read german.
but from the ones published online aces is the best, probably only followed by hardocp

[Indiana]

As saud even the review as it is now online is total BS or just made up.
There's absolutely no way they could gain that much in 3DMark (a single threaded app!!) due to HT being turned on. Just take a look at EVERY other review what you get in 3DMark2001SE with and without HT: nearly exactly the same result.

In the [L]ameOCP review they first had a nearly 100 points gain and now still have ~400 points gained. That just won't happen and if they really could back up their results why is every thread asking for this strange (and isolated to the HardOCP) phenomenon in their gorum immediately deleted?

I thought THG was bad, but this beats it by quite some margin...