Do you guys think AMD could've beaten Intel with Ryzen like, or was matching the

[Vishera]

I think it's an easy consensus that, after 15 years, AMD is finally back in the lead. Ryzen is pretty close, if not equal, to Intel's current offerings, for far less money. What I'm wondering is, if they were gonna undercut Intel anyways, wouldn't it have made sense to push the envelope a bit more to try and outperform them? Or did they make the right move here? Because it seems like, now that AMD has matched Intel in power, Intel has plenty of room to mop the proverbial floor with them with some amazing new processor line. What do you guys think?

 

[Bluefalcon13]

I think it's more of the process node they used. The architecture is good, and nice, but they used a low power design process (how they make silicon wafers into ICs with transistors on em). That's what is holding back the design more than anything from what I can tell. I'm interested to see some more power constrained options, such as laptops and such due to that reason. Would be interesting to see how they stack up against Intel in that regard.

 

[Vishera]

I think it's more of the process node they used. The architecture is good, and nice, but they used a low power design process (how they make silicon wafers into ICs with transistors on em). That's what is holding back the design more than anything from what I can tell. I'm interested to see some more power constrained options, such as laptops and such due to that reason. Would be interesting to see how they stack up against Intel in that regard.

I heard AMD is getting less than half yields on their wafers. Doesn't sound good.

What I see as a potential issue is if AMD had managed to make Ryzen 20% more powerful than Kaby Lake or Skylake, then coffee lake would have to be at least 30% more powerful to stay competitive. And Intel would probably go further than that. But by matching them, AMD has left Intel a ton of room to beat them with coffee lake and Cannon Lake/tiger lake/ice lake. There's no minimum requirement for Intel to shoot for to stay competitive.

 

[ShrimpBrime]

It's not all designed and aimed at desktop market. In fact desktop is probably the smaller profit of either manufacturer.

It's the Naples Server line you'd keep an eye on. https://www.amd.com/en/events/naples-tech-daym

 

[Woomack]

So far Ryzen is not beating Intel and is not even close. It's a good chip which is filling the gap between 2-4 cores and highly overpriced 6+ Intel cores but that's all. Ryzen is not making any special impression on most users and is dedicated to only small group of users. It was designed as highly efficient chip but entered market above its best efficiency range.
You have to remember that most users don't care about couple of % performance in some benchmarks while Intel platform offers more additional features, higher stability and better support. Most users are not rendering and don't need 6+ cores while in everything else any 4 cores are good. Most gamers are buying consoles, not new PC every 2 years.
Not to mention that many users decided on Kaby Lake because of delays in Ryzen premiere.
As I said, Ryzen is good chip but has a long way to beat anything from Intel as a whole platform.

 

[mackerel]

I generally agree with Woomack's post above. There are many ways to look at Ryzen, separately and as a system. There are good points, there are bad points, and most will ignore those if the price is right.

Agner Fog's recently updated architecture guide generally praises Zen, saying that per core per clock it generally has more potential than current Intel. I say potential, as it isn't manifest in single thread, and needs SMT to use that potential. This confirms earlier results showing cases where single thread IPC was below Intel, but it goes ahead with SMT. As we've suspected and later confirmed, the biggest weakness relative to Intel is in AVX2, with the usual "who needs it anyway" arguments coming into play. Ram bandwidth was also a concern, but this is in part mitigated by the large overall cache size.

IPC is one thing, but clocks are a "could do better" area for the enthusiast market. I'm choosing to operate mine at low OC (3.6) and low voltage to make use of that power efficiency, as the extra voltage and power to get gains beyond that add up rapidly. Personally I think the offering of select models around 4 GHz mark is enough to be competitive. Ok, it isn't the 5 GHz OC but that is a tiny niche.

Pricing is obviously a significant factor, and is in large part why in other tech forums I see a significant wave of recommendation to R5 for budget new builders. It will be even more interesting when R3 arrives.

But is that enough? For example, I saw on offer yesterday a laptop for £180-ish. It is only dual core Celeron, 4GB ram, 64GB SSD, and even comes with optical drive! For casual users, that's a whole computer for comparable cost to R5 CPU by itself. For a lot of casual users, that's enough. Is the gamer and other heavy lifting market enough?

I really hope Ryzen will lead to a reduction in Intel pricing, particularly in >4 core parts, but I'm not holding by breath. My other fear remains that by making Zen, let's call it "AVX lite", Intel may follow in future to keep cost competitive and we will start getting less features for our money.

On that note, the 1600 seems to have dropped enough in price since launch I might fill my 2nd mobo again and retire some i3 systems...

 

[Woomack]

Win for AMD would be to cover all that Intel series which have ridiculous prices but at the same time they have to provide stable platform. Releasing fixes for 3 months after premiere is for sure not helping and promising fixes in 2-3 months is not helping either. When AMD is talking about improvements and fixing things then it clearly means that they were not prepared for premiere. Many users are waiting for "final" ryzen as they don't want to be beta testers.
Next thing is that releasing next chipset and new socket for higher core ryzen is stupid move. Some users who got higher X370 boards are mad already. Some wait for new socket but there will be new Intel soon too. Somehow I feel like I purchased already dead socket and it's barely for 3 months on the market.

 

[mackerel]

I just ordered a 1600. I had to think for a while before doing that, as depending on use case, I know the similarly priced 7600k could be better. In the end, I did go 1600. While shopping I had noticed the B350 boards seem to have come down in price to something more reasonable, although X370 is still higher than I think they should be compared to Z270.

I always felt the current offerings do not offer complete competition against X99 and future series. AM4 is limited so if they were to enter that space, it would have to be a new socket. I don't see a limitation for consumer level use to stick on AM4 for quite a while. Maybe if AMD can do a Kaby Lake (compared to Skylake) with future Ryzen against today's, providing increased clocks through process optimisation, that'll cover many uses.

The platform stability is a mixed bag. I think the only significant problem is the ongoing ram situation. It may have improved, but it is still far from ideal. I've yet to see my system run ram beyond 2666 but I did manage that with 4 sticks.

Where are we going next with Ryzen? There's still R3 and APUs to look forward to. Not upgrades for sure, but they could provide compelling value solutions. I'm sure AMD don't want to be stuck in the low end but that is still something they have to address. Has there been an word on mobile offerings? Never mind server for now.

 

[Woomack]

I can't see any memory issue with Ryzen. AMD said officially what it supports and it works exactly as they said. Simply there is no issue and people are only creating stupid theories. Intel can work at higher clock but it doesn't change anything. The same was with previous series. Intel could run at higher clock but no one saw any issue with AMD memory frequency and now suddenly all need DDR4-4000 support or something isn't right ?

Big problem for AMD is that hardware manufacturers don't trust AMD. If they did then we would see more and better motherboards and better support for all devices. In theory you can buy cheaper motherboard and make budget ryzen build which will beat any intel in similar price but all those who know how it works won't even touch cheap amd motherboards as they are asking for trouble. Even higher series are just average.
APUs supposed to be cheaper series but again buying cheaper board you are asking for trouble. I don't expect to see any special graphics in lower APUs. Maybe the highest models will show something but knowing AMD buying high APU will be waste of money. Something that looks good in reviews and marketing stuff.

I haven't heard anything about new mobile amd chips. This will be really hard for amd to appear on the mobile market again. Right now they are not existing there, the same as on the server market. I don't think they can make anything that can compete with Intel in price. Maybe in price/performance if they deliver good chipset. They can try to make some 4-6 core mobile chips but I doubt that market will trust them enough to sell it in larger quantities.

I just see that AMD makes a lot of mistakes. They try to change the way how market sees them and push some new ideas ( which are not bad ) but in the end their marketing moves are pathetic the same as their premieres. Constant delays and constant issues look really bad and users are used to see AMD as not reliable platform ( regardless how it is in real ).

 

[Vishera]

I disagree on the APU argument. Even if the highest model is $200, as long as the iGPU is better than a 460 then it's still worth it, because Ryzen 3 plus a 470/570 or 1050/1050ti will still be over $200. That is, if the price is what we're expecting.

As for mobile, I doubt it's far off. AMD isn't gonna skip the mobile market, and I'm sure their APUs will appeal to mid range production laptop creators. Think about it, Pentium and i3 cover most of the $350-$600 market. They're only dual core processors. Then, in comes AMD with the A8 and/or A10 line, dual core and quad core CPUs with better integrated graphics.

Memory had definitely been an issue, support for JEDC is spotty at best, and they don't support the full speed of some super high end kits. I agree, not many people are buying those kits, but those that do might choose Intel instead now because they can't get the fullest from their money spent.

 

[ShrimpBrime]

Meh. 8 threads for 169$. I just purchased the 1400 for an HTPC build. 3 main components where 368$ My budget was 500$. Bought used GTX 770. Should be a decent build IMO. Got every thing else (needed) to complete the build, but went ahead and bought RAIJINTEK AENEAS case because I wanted to go a little smaller. Asus Prime A320M-K and Corsair vengeance LED PC24000.

Plug it in. Turn it on. Play games and build maps for Roblox is the main goal. Might even use it for some file serving as well.

Not a bad deal. At around that price I could buy the Intel i7 for 8 threads..... but it would sit on a shelf without a board the poor thing. Just way out of my price range.

HEY! I seen on Newegg the FX-9590 going for $129 on promo!

 

[wingman99]

I think AMD is doing the best they can with there engineers building Ryzen.

 

[ShrimpBrime]

I think AMD is doing the best they can with there engineers building Ryzen.

Stomps the previous line up. As they promised. That was the point. It was on point. FX and anything previous is lunch meat. 40%

 

[wingman99]

Stomps the previous line up. As they promised. That was the point. It was on point. FX and anything previous is lunch meat.

R&D builds AM4 first then they make a point of advertising. They did not best Intel, the engineers did the best they can.

 

[Woomack]

I disagree on the APU argument. Even if the highest model is $200, as long as the iGPU is better than a 460 then it's still worth it, because Ryzen 3 plus a 470/570 or 1050/1050ti will still be over $200. That is, if the price is what we're expecting.

As for mobile, I doubt it's far off. AMD isn't gonna skip the mobile market, and I'm sure their APUs will appeal to mid range production laptop creators. Think about it, Pentium and i3 cover most of the $350-$600 market. They're only dual core processors. Then, in comes AMD with the A8 and/or A10 line, dual core and quad core CPUs with better integrated graphics.

Memory had definitely been an issue, support for JEDC is spotty at best, and they don't support the full speed of some super high end kits. I agree, not many people are buying those kits, but those that do might choose Intel instead now because they can't get the fullest from their money spent.

I see no reason why anyone would spend money on APU and never use IGP which performance is not enough for new games. We were passing that couple of times on the forums. It's better to buy any lowest CPU + discrete graphics card. Highest APUs were always more expensive than lower CPUs. In theory all is great but in real who is using APU for gaming ? As many users as those who buy Intel CPU with IGP and play games on them. Literally those users who play games on IGP can play on any IGP regardless if it's high or low series. They're still too slow for new games at 1080p+ ( which is standard now ). Even if AMD IGP is 2-4x faster than Intel's then both are equally useless for games.

AMD is not existing on the mobile market, period. No one is selling AMD in higher quantities, there are no AMD professional/business series, there are no higher performance series. Current APUs in laptops have about 50% worse efficiency than Intels. About 95% laptops in stores are based on Intels. AMD has only older APUs. Current chips in laptops were designed 5 years ago+. So again AMD has a chance to appear on the mobile market as ryzen is highly efficient at 2-3GHz but it will be really hard to beat competition as users are used to Intel.
Even though Intel is offering almost only 2 cores then for most users it's enough and IGP in laptops have pathetic performance so APU doesn't change anything. Even Nvidia additional graphics are pointless. I see no point of some GPUs like GF920M which are not much faster than Intel/AMD IGP in games and additional features are almost the same.

All ryzen CPUs are compatible with JEDEC, all were working with JEDEC profiles since the beginning... where was the issue ? Every single memory kit which I tested on Ryzen was working as AMD declared, most even higher. There is no issue when memory works as it was designed. The only problems have users who wish to run high density modules at 3200 or any modules at 3200+ but who said it will work on this platform ? On X99 you won't run memory on most motherboards above 3200 too.
There is no official specification for 3200+ memory. All what is on the market right now is overclocked 2133-2666 and overclocking means it's not guaranteed. It can be tested on some platforms but no one will guarantee it will work on every motherboard.
No one is complaining when on Intel memory can't run above 3200. There are barely any motherboards which can make more than 3600+. There is only 1 IC right now which is in mass sales in memory kits above 3400.

 

[mackerel]

There has definitely been an improvement in ram support. With early bios my LPX 3000 2x4GB kit would only get as far as 2400. No boot at 2666. Not saying it was necessarily impossible if I had played around with enough timings and voltages, but it was far from plug and play. I can't recall which bios it was, but I got 2666 now. Not bad.

I've found out the hard way, high speed ram is still by no means guaranteed on Intel platforms either. Problems seem to start around 3000 and different kits like different mobos. On Ryzen so far, with only limited attempts, 2666 seems to be the sticking point with the jump to 2933 out of my reach. My ongoing criticism of Ryzen is the limited choice in ram speeds. I don't know if that is hard baked into the silicon but I'd love more steps.

 

[Woomack]

There has definitely been an improvement in ram support. With early bios my LPX 3000 2x4GB kit would only get as far as 2400. No boot at 2666. Not saying it was necessarily impossible if I had played around with enough timings and voltages, but it was far from plug and play. I can't recall which bios it was, but I got 2666 now. Not bad.

I've found out the hard way, high speed ram is still by no means guaranteed on Intel platforms either. Problems seem to start around 3000 and different kits like different mobos. On Ryzen so far, with only limited attempts, 2666 seems to be the sticking point with the jump to 2933 out of my reach. My ongoing criticism of Ryzen is the limited choice in ram speeds. I don't know if that is hard baked into the silicon but I'd love more steps.

Improvements were made by motherboard manufacturers, not AMD. It's just because motherboards were released without proper tests and we were beta testers for all available series. It shouldn't happen and it's what I'm talking about saying about lack of trust to AMD. Nearly all motherboards have better memory support now.

Intel Z series motherboards have almost always guaranteed 3200 memory clock. Everything above is marked as overclocking which depends on IMC and some other things. Personally I had no problems with memory clocks up to 3466 on any Z170/270 motherboard but there are clear differences between motherboards and some just won't run at 3600+ regardless what memory will be in use. Only top series will pass 3866 and I mean overclocking series with improved PCB. 4133+ can make almost only motherboards with shorter traces and 2 memory slots. There are single exceptions of 4 slot motherboards that can make 4000+ in dual channel ( on 2 memory modules, I'm not even mentioning 4 modules ).

New AGESA has been released but so far I haven't heard any news about memory compatibility.
Weird is that AMD recommended memory above 3200 when officially there are no boards which are supporting more than 3200. It just suggests that we can expect higher memory ratios in the future and some of us hope it will happen in this month.

 

[EarthDog]

Didnt read one post, including the first one...

No. If they could have, they would have. Lets put our thinking caps on before we post threads.

Edit: After reading the first post, lol, the only way amd is in the lead is if you use more cores, and want to save money. Most users, even here, do not benefit from more than 4c/8t..games wont (a lot of them so it males a difference) for years. Ipc, while vastly improved, is still not beating intel. Overclocking past their xfr is nearly nonexistent...

We have competition in the market which intel needs to respond to in some way... and that is good.

 

[Woomack]

I will give you some more time so maybe you read more posts ... I understand wall of text issue here

 

[EarthDog]

No thanks.