This is a split board - You can return to the Split List for other boards.

Why the hate against AMD?

#21KamenRiderBladePosted 9/17/2013 4:05:47 AM
OmegaDL50 posted...
I won't even click that link if that is the Tek Syndicate comparison, I'm just letting you know it's been proven across multiple forums and hardware that numbers they use are very dubious and inconsistent.

I don't see why you are trying to paint Intel in a negative light as a "business", and yet AMD does plenty of shady nonsense that can mislead buyers feeding their ignorance by ALSO doing their own rigging of benchmark scores to favor of chips to have usual performance differences in the synthetic numbers that don't even match the real world performance where it matters.

(I'm referring to this discussion here - http://www.overclock.net/t/1407562/guru3d-amd-fx-9590-centurion-benchmarks-surface with some of the numbers saying an 8350 is better then true Hex-Core 3930k and 3970x which anyone with some common sense would know is complete bunk.


The Tek Syndicate doesn't use synthetic benchmarks, they use games.

They were originally very much pro-intel too, then they tested the results themselves, and they found out the truth.
I'm not saying AMD win's every match, but they do win alot more benchmarks then Intel would have you believe.
---
Are you a MexiCAN or a MexiCAN'T - Johnny Depp 'Once Upon A Time in Mexico'
#22KamenRiderBladePosted 9/17/2013 4:08:56 AM
OmegaDL50 posted...
I won't even click that link if that is the Tek Syndicate comparison, I'm just letting you know it's been proven across multiple forums and hardware that numbers they use are very dubious and inconsistent.

I don't see why you are trying to paint Intel in a negative light as a "business", and yet AMD does plenty of shady nonsense that can mislead buyers feeding their ignorance by ALSO doing their own rigging of benchmark scores to favor of chips to have usual performance differences in the synthetic numbers that don't even match the real world performance where it matters.

(I'm referring to this discussion here - http://www.overclock.net/t/1407562/guru3d-amd-fx-9590-centurion-benchmarks-surface with some of the numbers saying an 8350 is better then true Hex-Core 3930k and 3970x which anyone with some common sense would know is complete bunk.


The thing is, I don't care about 9590, I don't plan on buying that one.

And I know how powerful the true Hex cores are, I'll be the first one to say how performant they are.

However, I'm not arguing about how good they are.

I'm arguing how AMD isn't as bad as Intel would like to paint them.

If Intel played fair and didn't write special code paths to check for Intel CPU's for their math library, then the performance gap between alot of benchmarks would be alot closer.

There are other people who have tested this theory and proven it to be true.

Intel has been prosecuted on this issue by the DoJ.
---
Are you a MexiCAN or a MexiCAN'T - Johnny Depp 'Once Upon A Time in Mexico'
#23ZeraphLordSPosted 9/17/2013 4:43:09 AM
Marikhen posted...
ZeraphLordS posted...
also amd is supplying the GPU and CPU of all the next-gen consoles, i'm not sure they're as pressed for money as you think; there are branches other than desktop computing


I seem to recall various statements over the years as to how console hardware is usually if not always sold at a loss. While it's not likely one has to wonder if part of that "loss" is passed on to component manufacturers. In the unlikely event that it is then AMD might not be getting as much money out of the deal as someone might think.


Agreed, but I have a hard time seeing AMD (or any company really) investing in an unprofitable venture (except to possibly deny their opposition). I'm of the opinion that AMD is doing fine for the forseeable future, and not because of their desktop offerings.

KamenRiderBlade posted...
If you actually bothered to watch the Youtube videos from the Tek Syndicate reviews

AMD does win quite a few benchmarks

Benchmark results can be cherry picked to show favorable results, it wouldn't be the first time in history that has happened


Video benchmarks? Hardly an efficient use of time; where are the graphs? Testing methodology is the most important part of any test, but I can't comment on them specifically. Relying on a single source is weak.

There are cases where AMD does win benchmarks, sure. I was commenting solely on the posted Anandtech benchmarks, which list any and all benchmarks done with the same hardware. Those games aren't cherrypicked, but the only ones performed where the CPU was the only variable, and thus the only ones comparable.

AMD has an advantage in multi-threaded apps, as I mentioned earlier. Whether that's significant or not is highly case-specific, because their brute power leaves something to be desired.
---
The best course of action is to just get the information you need, then get out while you're still alive. - destroy everything on GameFAQs
#24PraetorXynPosted 9/17/2013 5:24:10 AM
First and most importantly, you cannot directly compare these results to other benchmark results. The results in the video are for gaming WHILE RUNNING x264 encoding by using Xsplit to stream the game. So their testing is only relevant to gaming while x264 encoding a stream with Xsplit... If you don't game while running Xsplit, those results are meaningless.
Now who is tek syndicate? Minus points here. By doing a WHOIS on the domain, I see their domain was created in August 2011. They have existed for a little over a year, so I don't know any history on their credibility or what the background is for the people producing the content. For reviews, look somewhere with a history and blood in the game. It doesn't inspire confidence when the guy at Tek Syndicate ends the video with "too many numbers f***. Numberrrrrs. Ahhhhhhh". Ya, when doing reviews, you have to properly work with numbers - it isn't rocket science! And "f***" is simply unprofessional. In the bar, with friends, or even on Reddit, fine... When doing your job, not the time or place.
Next, their work is sloppy in their finished product. Sometimes in the first video he's saying one FPS, and the FPS presented on the screen is different. In their written results, they flip flop between presenting 1080 results first and 1440 results first - why? Because they can't keep their numbers straight. If they can't present things consistently in the final product they put all this work into, they probably made other mistakes before it was finished - it is anyone's guess what those mistakes were.
Why did they test with a top end AMD board against a mini-ITX Intel board? The Stinger is a fine board but this testing platform decision is illogical.
What they tested with (One high end board, a mini-ITX board, and a medium graphics card): FX-8350 w/ MSI 990FXA-GD80, 7870 GPU 3570k/3770k w/ EVGA Z77 Stinger mini-ITX Motherboard, 7870 GPU
What we tested with (all top of the line): 8350 w/ Crosshair V Formula, Asus 7970 DCU II TOP GPU 3770K w/ Maximus V Extreme, Asus 7970 DCU II TOP GPU
All our tests were done at 1080P, settings maxed. Straightforward and simple - it did not include their higher resolution testing. It did not include testing with Xsplit. So let's ignore 1440 because we don't have comparison numbers.
Take Metro 2033 for example. This is the best example that their test was erroneous for gaming performance (their test is only relevant for gaming while running Xsplit). After the results, they talk about how the 8350 kills in Metro 2033... WTF? If they were well informed they would be aware that at higher resolutions Metro 2033 is VERY GPU bound. Metro 2033 is a GPU bound bench, which means CPU power matters very little because the GPU is the bottleneck. But for whatever reason, when running Metro 2033 and Xsplit at the same time, their results on Intel were terrible (possibly because the FX-8350 has more cores to dedicate to encoding under Xsplit).
So the Metro 2033 results at 1080P? FX8350: Them - 36.44, Us - 36 3770K: Them - 27.48, Us - 37.3 3570K: Them - 21.2 3820: Them - 21.32
So on Metro 2033 at 1080P, they get WILDLY different results on the FX-8350 vs 3770K vs 3570K vs 3820. Something is wrong here. I don't know, but for instance, the 3820 is more similar to the 3770K in core/thread count, yet the 3820 results are more similar to the 3570K results... The 3820 and 3770K should perform similaly due to 4 cores/8 threads, while the 3570K only has 4 cores/4 threads. Their results just aren't logical. The test environment could have been bad.
I looked around, and similar criticisms to mine of these videos are on anandtech forums: http://forums.anandtech.com/showpost.php?p=34485701&postcount=14
Maybe read through that thread for further understanding.
Direct link to the question in the AMA.
I figured since it provided for some lively discussion I would repost it for visibility and further discussion.

---
Console war in a nutshell:
http://imgur.com/xA6GJZ9
#25TimePharaohPosted 9/17/2013 7:11:05 AM
d209999 posted...
Because AMD = FINISHED

It is known.

---
"HE are genius, firstly." - ASlaveObeys
http://i.imgur.com/SQAc17B.png
#26ChocobonerPosted 9/17/2013 7:23:16 AM
Really there is a lot of hate on AMD especially around these forums, it always leads to "why buy the 8350 when you can get the new haswell for 20 bucks more and get much better performance". Yeah, you can. Or you can buy the 8320 for $140 bucks, slap on a 25 dollar cooler, OC it to 4.3-4.4 and get the same performance as a stock i5 haswell give or take 5%. You're not missing out on much if you opt for intel over AMD, especially when it comes to gaming it's a much wiser option to take that 60-70 bucks you shaved off and get a better GPU. Who knows what next gen gaming has in store, IIRC even developers were saying people should opt for AMD cpus in the near future. Unless we're talking about streaming, in which case AMD takes the cake.
Found a forum link with quotes, no time to look for original http://linustechtips.com/main/topic/10522-game-developers-choose-amd-over-intel-for-gaming/
#27CC RicersPosted 9/17/2013 7:51:31 AM
Kilron posted...
Supposedly AMD cards don't handle antialiasing as well as other cards. Not that I would know, I never use AA. I do know some programs like Blender simply don't work as well as they should or are flat out unable to use some features, but if what I'm reading on their support forums is true, AMD is working on better Blender support as we speak.


I have an aging HD 4670 and it does fine with AA. Its Catalyst utility also let's you override AA settings, great for games that don't have that option, like Mass Effect.

I also use Blender a lot and haven't seen many problems or glitches with their 3D views. Maybe they're talking about the new previews for the Cycles engine.
---
WikiLeaks scandal: WikiLeaks is not a true Wiki!