Why is HardOCP's numbers so far off?

I also notice that the AMD power use is very close on both graphs, while Conroe on HardOCP is boosted up severely...

I find it very odd that 'turning off power saving features' would limit the intel chips, but not the AMD ones... I'm not gonna jump to conclusions, it could just be a huge coincidence, but... it's all very, very odd...
 
Yeah, those numbers seem way off compare to most other reviews I have seen, at least under idle...the 148W is about what everyone else is reporting...Load power usage is about what people are seeing though...in the low 200s.
very interesting...
 
And if the numbers are true, then perhaps Conroe was made for energy saving features? Esp. since sites that use them get signifigantly less power usage. We'd like to see a new benchmark with Conroe running games using energy saving features, and how it runs at 100%...
 
Newsboys2004 said:
And if the numbers are true, then perhaps Conroe was made for energy saving features? Esp. since sites that use them get signifigantly less power usage. We'd like to see a new benchmark with Conroe running games using energy saving features, and how it runs at 100%...
Click to expand...

It looks like on power consumption:

Hard OCP didn't enable speed step which makes it s idle consumption about the same as an equally clocked FX.

Anandtech did enable it since all 3 processors had exactly the same idle consumption (b/c they were all at 1.6Ghz = 6x multi), but didn't mention it.

Tom's Hardware did enable it, and was thorough enough to mention that they did enable it, and how to enable it, if you happen to have one.

Then there could also be other factors/variables at play, like PSU, other system components, etc..

Then on gaming:

The benchmarks on HardOCP weren't all apples to apples, that might be part of the reason why they're not showing as big of a difference. Also Anandtech used crossfire + apples to apples, which let the CPU's stretch their legs a little more and show off more of a difference. Then Tom's used a lower resolution which would be less GPU bound, which showed a little more difference.


I liked Anand's gaming benchmarks b/c of the crossfire. I also like Tom's power consumption benchmarks b/c he showed all 3 states, load, idle & speedstep and explained the difference a little more thoroughly.

Anyway, time for everyone to go buy a new PC. :) j/k
 
The numbers don't make sense because they were never intended to. This is an obvious attempt to garner attention and traffic for a website. Good Luck with your future reviews, because without credability, you really don't have much.
 
cronic007 said:
The numbers don't make sense because they were never intended to. This is an obvious attempt to garner attention and traffic for a website. Good Luck with your future reviews, because without credability, you really don't have much.
Click to expand...

To whom are you referring? I've always been able to trust {H]'s reviews. They don't have to generate traffic because it comes from their reputation as an honest review site that doesn't post bullshit information to make themselves look good.
 
Lol well there's 2 ways to look at this:

1) Nobody plays at 640x480, or 800x600...unless you got a CRT still (than you can).

2) How many people have CF 1900s or SLI 7900GTXs? I'm betting less than 0.5% of the gaming population. I'm betting more likely that 7900GT (single card) rule the roost...thus making those benches helpful...but again, not very relevant.

Thus the only real benches that matter are 1024x768 and higher. The 16x12 resolution represents your typical 20" 4:3's...and the 19x12 for the many 2405 owners. Yeah we could play lower resolutions...but I doubt many LCD owners deviate too much from their native resolutions.

Intel has always been strong in encoding/mp3 ripping/etc, even with their Netburst they were already on par or better--so no surprise there.
 
ryan_975 said:
To whom are you referring? I've always been able to trust {H]'s reviews. They don't have to generate traffic because it comes from their reputation as an honest review site that doesn't post bullshit information to make themselves look good.
Click to expand...

Exactly. They aren't going to say "go out and buy a Conroe now because you'll get framerates that are twice as high in every game!", then tell them "oh, we forgot to mention, to get the 2x framerates you have to buy two 7900GTXs and then game at 800x600 so your game won't be GPU-bound" when someone wonders why they are seeing no difference in their game's framerates.

Instead, they're going to be honest and let single-card users know that upgrading to Conroe may not improve their framerates as much as they think it will.

Also, they tried to use SLI. They had problems with their GX2, and it's not like they could just suddenly change everything to Crossfire either. Intel didn't give them enough time to do that.
 
Mav451 said:
2) How many people have CF 1900s or SLI 7900GTXs? I'm betting less than 0.5% of the gaming population. I'm betting more likely that 7900GT (single card) rule the roost...thus making those benches helpful...but again, not very relevant.
Click to expand...

Crossfire is relavant looking forward, b/c the next gen cards will likely be as powerful as 2 last gen cards in crossfire. So its best to not let the tests be GPU bound if possible to showcase the true power of a CPU.
 
Mav451 said:
Lol well there's 2 ways to look at this:

1) Nobody plays at 640x480, or 800x600...unless you got a CRT still (than you can).

2) How many people have CF 1900s or SLI 7900GTXs? I'm betting less than 0.5% of the gaming population. I'm betting more likely that 7900GT (single card) rule the roost...thus making those benches helpful...but again, not very relevant.

Thus the only real benches that matter are 1024x768 and higher. The 16x12 resolution represents your typical 20" 4:3's...and the 19x12 for the many 2405 owners. Yeah we could play lower resolutions...but I doubt many LCD owners deviate too much from their native resolutions.

Intel has always been strong in encoding/mp3 ripping/etc, even with their Netburst they were already on par or better--so no surprise there.
Click to expand...

Everyone who argues this point, listen up. If he knew that the GRAPHICS CARD would bottleneck, why would the reviewer continue to use those resolutions unless he wanted to prove something else. I don't trust HardOCP reviews anymore. If the reviewer acknowledged that his single card solution bottlenecked, why would the reviewer continue to test after knowing this bottleneck. I've yet to see any truely credible review site openly acknowledge that the graphics card were bottlenecking and STILL continued with those resolutions. That's a bullshit way of reviewing a CPU.
 
they did it at low resolutions to take the GPU out of the equation. THis is a CPU reviiew. Higher resolutions would cause the GPU to be a bottle neck and wouldn't show the true results of a CPU comparison. And it also says right in their review that there were going to do a SLI based review but the motherboard they had for Conroe wasn't able to run the 7950. But the 7900GT(X?) that they did use isn't exactly a chimp either.
 
StealthyFish said:
That's a bullshit way of reviewing a CPU.
Click to expand...

Well, if you read it was NOT a CPU review. We looked at the real world gameplay it could deliver with a top end video card. Who told you that was a CPU review?
 
If they can't read the first time, what makes you think they can read the second? :)
 
Hehe, that won't make any difference. :)

Got my E6700 up to 3.15GHz / 315 CPU at stock voltage......next stop, 3.5...maybe. ;)
 
Seiously though, it has been a fun day. I like to argue, so I had a blast. I was comfortable with my position, although I did go through some soul searching about 4am, but I was done by breakfast and had regained my confidence. Not a bad thing to question yourself and your motives, just sucks if you find out they are wrong. ;)
 
not as bad as when you (generically, not specifically) keep going thinking they're right and never know they're wrong.
 
I like HardOcp reviews. I just wish in this case that they would have used 1280x1024 4xAA 16xAF also.

I understand where Kyle is coming from. Basically if you play at these uber graphics settings dont expect to gain many FPS with a new conroe.

The problem is the settings he has chosen are ridiculous if you try to play games like BF2 competitively. I mean look at his FPS chart its all over the place 80 one second 30 the next. This causes the game to feel really unpredictable to me. Thats why even though I have a pretty nice rig I run at the settings I posted above.

So my main questions about Conroe relate to BF2 and the settings I posted above.

1. Is this the processor that will finally allow me to pretty much stay pegged in the upper 80's the whole time with no dips? Like I get now in maps like Sharqi and Daqing Oilfields.
2. Will this processor improve the experience in 50+ player servers where the performance is noticeably worse than in say a 32 player server. Or is it simply a server lag issue.

I would be interested in hearing anyones opinion on these two questions.

Thanks
 
SigmaOrionis said:
1. Is this the processor that will finally allow me to pretty much stay pegged in the upper 80's the whole time with no dips? Like I get now in maps like Sharqi and Daqing Oilfields.
Click to expand...

There is pretty much no process or that will solve problems like these. Alot of this has to do with how the map is designed and how the engine or graphics drivers deal with textures in the distance you can't really see. Back in the Quake3 days some people's custom maps that weren't all that complex would bring a system to its knees at the times, because the engine tried to draw absolutely every triangle without accounting for view distance. Later games based on the Q3 engine (MOHA, BF1942) overcame some of these problems in the software, without requiring a massive hardware upgrade.

SigmaOrionis said:
2. Will this processor improve the experience in 50+ player servers where the performance is noticeably worse than in say a 32 player server. Or is it simply a server lag issue.
Click to expand...

Yeah, if the server is having problems with lag due only to CPU issues, of course Conroe will improve this problem. But remember, server lag can come from bandwidth, memory and CPU loading, so Conroe in and of itself isn't the answer here.

I was fortunate to sit in on a lecture from one of the intel R&D engineers, and the lecture was covering what teraflop single processor computing could bring to the world. The short of it is that every time new hardware comes up, developers are excited because they can finally write the software they have been dying to develop but never could since hardware limitations prevented it from being a viable solution. I remember the quote "if we had 64 bit available to us 3 years ago, we'd have been coding in it back then already," implying that alot of software is built around what the expected hardware limitations will be. I mean, if you're developing a new game engine, you don't want to make it so intensive that it can only run on the absolute fastest machines in existence. If you do, you're not going to sell many copies until the mean hardware platform catches up.

The bottom line is that game specific (particularly map specific) problems are not always hardware limited, but sometimes can be optimization issues in software, not just the game but the hardware drivers as well. As a result, there is no "magic bullet" piece of hardware that will solve every issues with performance. The "system" is a conglomerate of different parts that must effectively work together.
 
look, H is about gaming... and the review was showing that if you game and have a high end AMD system already, there is no need to chuck it and buy a whole new system yet.

Why are people hating on the H reviews..?

Sure, conroe is fast... but the money, right now, is better spent on a good video card, if you are a gamer.

Plain and simple.

HardOCP is never afraid to tell it like it is... Most review sites will NEVER ever EVER give a bad review... they don't want companies to stop sending review items.

H goes out of their way to even buy products to review them... they are not afraid to say that, while conroe is fast, there is just no need to chuck your high end AMD system yet.

Nobody games at 640X480, and if that is the only place you see conroe advantage in games, it is kindof sad that most review sites don't point that out.

I am fucking glad that H tested at resolutions that *I* actually use... Those numbers are far more important to me if I was to be considering putting together a new system.

now bring on the 64bit tests... it is interesting that we have not seen any...
 
I'm glad they did this too. I have an Opteron 165 at 2817MHz, basically FX-62 speeds. I play games at 1680x1050 and use a 7900GTX. Now I'm totally fine without having a Core 2 system, since my system is already just as fast in games. If I was doing more DVD ripping, then I might want a Core 2. But I can just play games with the dvd rip in the background anyway. I guess if you were on Socket A/754/478/775 (915/925) then Core 2 would be a worthwhile upgrade for gaming, but PCIe is the most important thing to have at this point.
 
The bottom line is this... everyone agrees that conroe owns. They have the best performance/price ratio AND the best performance.
 
yeah conroe pwns... however, gamers have better facilities on socket AM2 and 939 at the moment.

better SLI boards... better crossfire boards... better overclocking boards.

right now this is true.

Maybe after conroe has been out a few months and more boards start trickling down the line... things will change... DFI is building a p965 board... things like that.

See, on AMD, the best chipset is nvidia nforce. On intel the best chipset is intel's own... so there is a big support difference there with regards to the enthusiast community.

Personally I don't mind having a few less options, but still have stability and speed of intel. Though, I would miss the out-of-the box tweaking options that are standard on alot of AMD boards.
 
You must log in or register to reply here.
Back
Top