sunaiac wrote:

Except the 8350 is faster than the 2500K in highly multithreaded applications. All the ones tested when the CPU came. All of them.

Well no. Just look at the Anandtech article I linked:
http://www.anandtech.com/show/6396/the-visher … fx4300-tested/3
http://www.anandtech.com/show/6396/the-visher … fx4300-tested/4
Compiling is a highly multithreaded scenario where the FX8350 is losing for example.
It also loses in Photoshop, which again uses multithreaded filters.
I'm sorry reality doesn't agree with you.
Only highly synthetic multithreaded benchmarks can make FX8350 outperform the i5 2500. Real-world multithreaded benchmarks such as Photoshop, Visual Studio or indeed most modern games rely on more than just a bunch of cores running a bunch of threads.

sunaiac wrote:

But I'm very happy to learn that a core can run multiple threads. Who'd have thought ?

Well if you had just read my wellknown blog about that, you may have learnt something: https://scalibq.wordpress.com/2012/06/01/mult … ulti-threading/
I'm just tired of the same old clueless arguments over and over again.

Besides, the argument is rather pointless in 2015. Intel has updated the i5 a few times, while AMD is still selling the FX8350. The i5 2500K is no longer its competitor, it's up against faster CPUs now.

It's very hard to judge the performance of theses FX processors unless a game has code for the FX module it may run slower. Usually I always go with Intel and I am not saying one is better than the other, AMD vs. Intel the never-ending debate, but my FX8120 at 3.9Ghz with 2133Mhz ram has kept me happy. Also I am surprised no one has mentioned that AMD purchased RISC from Cisco and used that tech to design theses CPUs.

Nice mid-low end gaming PC.

alexanrs wrote:

This probably reaches the performance of an i5 for gaming, I'm just curious to know which.

i5 and i7 are equal in modern multiplatform games, wich are made for outdated PS4 crap. Current i5 may to be faster in some games than AMD, - in 1-4 cores tasks. Also i5 is better in DOSBox.

Thanks for all of your inputs. Yeah, I do realize that what I have isn't exactly the best, although I was pretty much going for a budget PC that can play today's game just fine. Sure, it is not a high-end core i5, or an i7, but it does the job for now. Maybe my next one is going to be a core i10 or i12. (Maybe, 🤣!)

philscomputerlab wrote:

Come on, Mike just built his new PC, let him be and enjoy it 😀

Thank you! I appreciate the sentiment! 😀

When the crapdozer first came out the disappointment was real and ever since never bothered with AMD in the high end again. AMD should have stayed with Stars from the first APU line and improved it from there. The L2 and L3 performance as well sharing the FPU along with too many other things between cores was horrendous.

Keep in mind that some benchmarks are heavily Intel optimized, and/or will ignore AMD's specialized SIMD. ICC compiler for instance for some libraries will optimize code for the latest Intel SIMD but if the CPU is not Intel it will default for older SIMD or in some specific cases run 386 code. This has been happening for years and specially important for Intel during the Pentium 4 years because: A) It made Intel CPU's look faster than they really were; B) It crippled AMD 's CPU's performance on benchmarks to avoid the obvious performance gap.

It is also a fact that:
Intel bribed OEM's to not buy AMD thus preventing AMD to gain significant market share;
Japan Fair Trade Commision investigations concluded Intel was doing these shady practices;
Intel and benchmark software producers were crippling AMD performance on purpose;
Michael Dell lied in court when testifying in Intel's favour;
Intel's ICC compiler still does unfair dispatching;
It is possible via scripts to improve AMD's benchmark results by removing "cripple AMD" function;

There's no doubt that AMD Bulldozer architecture is anemic (which was further exarcebated by Intel's reluctance in adopting SSE5 BTW) compared to Intel's comptemporary but the difference is lower than what we are led to believe by the industry. Intel has the bread and butter, they call the shots and they will never repeat the Netburst mistake that allowed for a competitor to gain market share (and could have gained SIGNIFICANTLY more) and threaten their dominance.

So keep in mind when some say 'AMD sucks' or 'Crapdozer' that your opinions might be a tad exxagerated.

I hope that Zen CPUs will put AMD back in the game. 😎

carlostex wrote:

Keep in mind that some benchmarks are heavily Intel optimized, and/or will ignore AMD's specialized SIMD. ICC compiler for insta […]
Show full quote

Keep in mind that some benchmarks are heavily Intel optimized, and/or will ignore AMD's specialized SIMD. ICC compiler for instance for some libraries will optimize code for the latest Intel SIMD but if the CPU is not Intel it will default for older SIMD or in some specific cases run 386 code. This has been happening for years and specially important for Intel during the Pentium 4 years because: A) It made Intel CPU's look faster than they really were; B) It crippled AMD 's CPU's performance on benchmarks to avoid the obvious performance gap.

It is also a fact that:
Intel bribed OEM's to not buy AMD thus preventing AMD to gain significant market share;
Japan Fair Trade Commision investigations concluded Intel was doing these shady practices;
Intel and benchmark software producers were crippling AMD performance on purpose;
Michael Dell lied in court when testifying in Intel's favour;
Intel's ICC compiler still does unfair dispatching;
It is possible via scripts to improve AMD's benchmark results by removing "cripple AMD" function;

There's no doubt that AMD Bulldozer architecture is anemic (which was further exarcebated by Intel's reluctance in adopting SSE5 BTW) compared to Intel's comptemporary but the difference is lower than what we are led to believe by the industry. Intel has the bread and butter, they call the shots and they will never repeat the Netburst mistake that allowed for a competitor to gain market share (and could have gained SIGNIFICANTLY more) and threaten their dominance.

So keep in mind when some say 'AMD sucks' or 'Crapdozer' that your opinions might be a tad exxagerated.

I couldn't agree more!
In my main PC I have a 2500K. And a E8400 before that.

But, in other applications, I choose to support AMD. My homeserver has an Athlon X4 840 FM2+ (no need for display output) and the "movie watch PC" is an AM1 Athlon 5350 powered with my laptop's brick. 🤣
The homeserver motherboard was the only motherboard I could find new to support RAID and have 8 onboard angled sata connectors for ~80 euros and the AM1 drains almost no power at all, yet it managed FHD at any bitrate thanks to the Radeon IGP.

I really don't give a crap about either AMD or Intel. What matters is that this whole debacle unfairly hurt AMD the most, but it also affected the consumers. AMD also put themselves into trouble with incompetent CEO's like Hector Ruiz, specially the ATI deal. AMD wanted to buy NVidia instead, but apparently Jen-Hsun Huang was asking too much (or maybe history proved he wasn't) as he wanted to become CEO of the new merged company. Hector had too much weight in the board of directors and the decision was to get ATI instead.

carlostex wrote:

I really don't give a crap about either AMD or Intel. What matters is that this whole debacle unfairly hurt AMD the most, but it also affected the consumers. AMD also put themselves into trouble with incompetent CEO's like Hector Ruiz, specially the ATI deal. AMD wanted to buy NVidia instead, but apparently Jen-Hsun Huang was asking too much (or maybe history proved he wasn't) as he wanted to become CEO of the new merged company. Hector had too much weight in the board of directors and the decision was to get ATI instead.

Well put.

True, Intel did some shady things, but AMD shot itself in the foot.

Two years after Core2 was rocking the mobile world, AMD is puttering about with their Kuma design. Oh whoopee! Hybrid crossfire! Slower than just using the discrete by itself, but the gimmick is what counts.

carlostex wrote:

There's no doubt that AMD Bulldozer architecture is anemic (which was further exarcebated by Intel's reluctance in adopting SSE5 BTW) compared to Intel's comptemporary but the difference is lower than what we are led to believe by the industry. Intel has the bread and butter, they call the shots and they will never repeat the Netburst mistake that allowed for a competitor to gain market share (and could have gained SIGNIFICANTLY more) and threaten their dominance.

I still can't believe AMD even released Bulldozer. Technically it is not that bad, but they had no leverage to allow it to take off. Their CMT (clustered multi-thread) doesn't shine with code optimized for good old SMT, they don't have their own compiler to really extract its full potential, and neither the marketshare and influence to really get a large chunk of the industry to really optimize their code for it. I hope Zen rocks! AMD processors served me very well in the past, and I hold no PC I have ever owned as dear as my original Duron! That thing allowed a teenager from a mid-low class family like me to finally have a decently fast PC for gaming! I'd love to consider AMD processors again when building a new PC... as it stands now, anything i can get from AMD is a downgrade even compared to my secondary PC (Sandy Bridge i7).

stuvize wrote:

It's very hard to judge the performance of theses FX processors unless a game has code for the FX module it may run slower.

The problem is that when 4 out of 5 machines are not based on the FX-architecture, there isn't much incentive to develop for this.
The Pentium 4 found itself in exactly the same situation (and it baffles me that AMD even tried): with properly optimized code, the P4 could perform quite well. But it was quite weak at all the code that was already out there, which was optimized for Pentium 3 and earlier CPUs.
In the x86-world, performance on legacy code is very important.

alexanrs wrote:

I still can't believe AMD even released Bulldozer. Technically it is not that bad, but they had no leverage to allow it to take off. Their CMT (clustered multi-thread) doesn't shine with code optimized for good old SMT, they don't have their own compiler to really extract its full potential, and neither the marketshare and influence to really get a large chunk of the industry to really optimize their code for it. I hope Zen rocks! AMD processors served me very well in the past, and I hold no PC I have ever owned as dear as my original Duron! That thing allowed a teenager from a mid-low class family like me to finally have a decently fast PC for gaming! I'd love to consider AMD processors again when building a new PC... as it stands now, anything i can get from AMD is a downgrade even compared to my secondary PC (Sandy Bridge i7).

I didn't want to get into this discussion, but I can't help myself 🤣

Yeah, I wonder why they even made the decision to design Bulldozer the way it is.

As for me, I've had AMD processors in my main PC from '92 to '11 (except for a brief UMC episode), then went Sandy Bridge because AMD had nothing to offer and still hasn't.

But it's good to see people like the OP holding the AMD flag up!

oerk wrote:

Yeah, I wonder why they even made the decision to design Bulldozer the way it is.

Yup, something went horribly wrong there.
As an assembly programmer, I have quite a good idea of what a CPU should be capable of in terms of pipelining, instruction-level parallelism, throughput etc.
And I have experience optimizing for both Intel and AMD CPUs... So when I saw the first details of the Bulldozer architecture, I thought "Wait a minute, that isn't going to work... they're removing some execution units per core, which will greatly cut into the performance".
Then John Fruehe started interacting on forums everywhere, making all sorts of claims that based on the specs known so far, would be completely unrealistic. Because they were going from 3 to 2 execution units per core in some cases, claims of cores having higher IPC would mean that these execution units would have to work more than 33% faster.

I already had a bad taste in my mouth from a few years earlier, with AMD's Barcelona, where they made claims of 'native quadcore' and '40% faster than any x86 on the market'... It wasn't 40% faster. It wasn't even 1% faster. It was SLOWER.
So this time I put up a blog, addressing Fruehe's claims, more than a year before release. I *knew* AMD couldn't pull it off. I caught so much flack for sticking my neck out like that. You wouldn't believe all the rabid AMD fanboys out there, desperately believing Fruehe's claims and defending them with the most horrible flames.
But in the end, obviously I knew exactly what I was talking about, and everything I said turned out to be true. In fact, Bulldozer was even worse than what I predicted.
The blog is still there (John Fruehe is not): https://scalibq.wordpress.com/2010/08/3 ... test-liar/

I guess what I'm saying is: if I knew Bulldozer wasn't going to work, and I could point out exactly *where* it wasn't going to work, more than a year before release... Why didn't the people at AMD see this, or if they did, why didn't they take action? Bulldozer should have been abandoned and Stars should have been refreshed instead. From here it looks like AMD has absolutely no idea what they're doing.

Zen looks promising, as it is right now the FX series make great budget gaming machines. I use a FX-8350/HD7870 system and absolutely love it. Great on the budget (my card was $100) and I can not feel the difference between it and my friend's i5-3570k/Geforce 770 setup in everyday tasks and even in most games.

Zen will not and does not need to beat Intel. It just needs to be a cheaper and good enough alternative. If they reach Sandy Bridge IPC levels, its good enough for me.

carlostex wrote:

Zen will not and does not need to beat Intel. It just needs to be a cheaper and good enough alternative. If they reach Sandy Bridge IPC levels, its good enough for me.

It needs to perform well enough for AMD to be able to sell their processors without constantly dropping prices of their products to remain competitive. Those price drops eat away their profit margin and that is terrible for AMD in the long run. I really doubt the current FX processors are much cheaper to make compared to Intel offerings, with the difference that Intel can sell at a higher price and, therefore, have more money to invest.

alexanrs wrote:

It needs to perform well enough for AMD to be able to sell their processors without constantly dropping prices of their products to remain competitive. Those price drops eat away their profit margin and that is terrible for AMD in the long run.

Yup, that's the thing... AMD knows they can't beat Intel in terms of value. They've been bleeding the company dry: http://arstechnica.com/tech-policy/2015/10/am … -straight-loss/

The moment that AMD took over ATi, they were fighting a war on two fronts: Intel and nVidia. It seemed inevitable that poor results in one division would eat into the other division as well at some point.
It seems we have arrived there now... AMD's DX12 lineup was little more than a rehash of their 3-year old GPUs, with HBM being the only new addition. Unlike Intel and nVidia, they do not support the new rendering features in the DX12 API. Also, they marketed the Nano, which is supposed to be a high-end HTPC card, for 4k... but because of their outdated tech, they lack the HDMI 2.0 support that you need for 4k TVs...

Guys, this is getting insulting. Mike just got a new computer that he apperantely likes. I suggest you hijack badmojo's 286 or retrofanatic's Turbo XT thread and make fun of them because those two do not run Crysis.

Intel vs. AMD banter is so outdated and stale. Everything worth mentioning was pointed out a few months after Bulldozer saw the market. All that is left is a pointless fanboy quarrel.