heatlesssun
Extremely [H]
- Joined
- Nov 5, 2005
- Messages
- 44,154
I agree. But AMD really needs to have something that can rule at the top end, period.
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
AMD Ryzen 5 Processors Start At $169 and Launch on April 11th
- Thread starter Zarathustra[H]
- Start date
I agree. But AMD really needs to have something that can rule at the top end, period.
Nope. They only have to get close enough that on most work loads the difference doesn't matter much while also costing significantly less. Which they have achieved. You might not like that strategy but it worked well for them in the K6 days and it will work well now.
Personally I prefer it this way since I'm a bang for the buck type guy. Whenever they have the top end chip they charge just as much as Intel.
What amd has is good enough at the top. With the R&D and manpower they had. They did good enough. They beat the matched intels 1000 dollar chip with 500 dollar one. Yea not as fast in gaming but I doubt people buy 8 core chips to game all day long. It did well where it mattered. As long as we know IPC Is there clock speeds with come as the process and architecture matures. Thats nothing to worry about.
heatlesssun
Extremely [H]
- Joined
- Nov 5, 2005
- Messages
- 44,154
What AMD needs more than ever is another Athlon. Regardless of cost AMD just needs parts people lust for. Ryzen is certain a good part especially for the price, but it's not lust worthy.
What makes it lust worthy is amd has sucked so hard for so long that they got even this close to intel. Intel has has no competition for years and now they do. I will take that, this will probably turn in to an Athlon down the road with a few tweaks. They have a great baseline to build on.
Only some PC enthusiasts lust over PC parts, the rest of the world cares mostly about performance vs cost, and the rest of the world still buys more PC's than enthusiasts. Also quite a few PC enthusiasts do care about bang vs buck too. That is how overclocking really got started anyways you know. So its quite valid and financially sound for AMD to target the bang vs buck with Zen.
Realistically given their limited resources and the non-Intel fabs they have to use Ryzen is about as much as you could hope from AMD right now. Ryzen 8 has a shot at taking the performance crown...at stock clocks anyways. If you really, personally, are only interested in a peak Intel-beating performance chip from AMD you're stuck waiting for that.
Your Microcenter sucks?
http://imgur.com/a/cl2H5Ryzen was only supposed to be able to keep up with Broadwell-E pre-launch, so I'd say they're off to a pretty good start. It's not a wet dream for [H] readers, but like somebody mentioned a page or two back if the R5's at stock are less than a Core i5 and perform on par with the i5 at a lower price (or perform somewhere between an i5 and i7 because of the extra cores/threads), then you're looking at being able to get a beefier GPU on the same budget. If I'm spec'ing out a gaming rig for someone else I don't really want to support overclocking. I want it to be zippy and problem free.
Benchmarks using cheap 1080p60 monitors are going to be relevant for R5 reviews because that's what a lot of people spending $500 or less are using on their gaming rig.
Tsumi
[H]F Junkie
- Joined
- Mar 18, 2010
- Messages
- 13,752
It only matches it at stock vs stock in CPU performance. The i7 overclocks much higher, has more PCI-E lanes, and more memory channels. $500 is a reasonable price for what Ryzen is, and by no means a killer of the $1000 chip. The 1700 on the other hand… that is the chip that will really shake things up.
That is his Microcenter. I searched it using his original link. I have no Microcenter in my state. Or Fry's either.Your Microcenter sucks?
Also 4 mobo's is not a lot.
Which is what will matter for the most part.
No on really cares about the extra PCIe lanes or memory channels for X99 though. Neither matter much for most things outside of HPC or server style work loads. For Ryzen most of the gains from the higher clocked DDR4 RAM are actually from the IMC and IF getting clocked higher as a result of the faster RAM and not necessarily a main memory bandwidth issue.
10 core intel chip doesn't overclock that high either anyways. I was talking about the rumored x399 chipset even if it exists. obviously that will have more lanes from what it looks like. (Thats all up in the air and taking it with a grain of salt). Most people in desktop environment dont really use all those lanes anyways. I have an x99, rocking a single high end video card here lol.
KazeoHin
[H]F Junkie
- Joined
- Sep 7, 2011
- Messages
- 9,000
The problem is that the average, bang-for-buck, Joe buys his PC from an enthusiast. And that enthusiast's bias plays a huge role in what hardware they suggest. This is why companies with top-performance parts are always the dominant market leader despite others having better performing and cheaper mid-range parts.
LOL no. They buy Dell's or Lenovo's. Very few people go custom build or to an enthusiast.
True, but I was kind of expecting 1080-at-launch levels of availability before I checked. I just upgraded, so I can't justify buying a Ryzen chip, but it's nice to see they're available. The ASUS Prime boards (they were renamed that) is the series I've been going with most upgrade cycles, so seeing it totally taunts me.
Singularity_Survivor
Limp Gawd
- Joined
- Apr 7, 2016
- Messages
- 212
Terrible post man, Intel only officially supports DDR4 2400 on the 7700K yet you post junk about Ryzen's official DDR4 2666 spec. Kyle, you might want to pull his Staff Member banner. He's clearly biased or lacks the knowledge needed to be on the [H] Staff team.
![Zarathustra[H]](https://cdn.hardforum.com/data/avatars/m/9/9298.jpg?1707758471){kind=avatar}
Zarathustra[H]
Extremely [H]
- Joined
- Oct 29, 2000
- Messages
- 38,830
I think you missed the point I was trying to make.
AMD, in their blog seem to suggest runnning DDR4-2933
AMD's max specs say DDR4-2666.
This has nothing to do with Intel.
AMD has a performance issue and they are recommending you solve it by running outside of their official specs. I found that amusing.
Singularity_Survivor
Limp Gawd
- Joined
- Apr 7, 2016
- Messages
- 212
Either way, you even admitted you didn't know that Intel only officially supports DDR4 2400 for their latest top performance 4/8 cpu, 7700K......You should know that kind of basic information.
Anyway, I like that AMD is officially recommending overclocking, I find it very [H].
Tsumi
[H]F Junkie
- Joined
- Mar 18, 2010
- Messages
- 13,752
The point is the price difference is there because there are significant differences between the CPUs. Whether or not the difference is worth it is on an individual basis.
With the PCI-E controller being integrated on the CPU on all modern CPUs, there is no magical way to add GPU dedicated PCI-E lanes by chamging the chipset.
Edit: NVM, x399 will be a completely different socket if it exists.
People with enthusiasts in the immediate family use computers built by the enthusiasts unless it is a laptop or tablet. Others buy based on input from friends, family, and salespeople, who may or may not be informed.
Last edited:
KazeoHin
[H]F Junkie
- Joined
- Sep 7, 2011
- Messages
- 9,000
Not true. People who want a good PC often buy from a guy like me (also, 8 years experience as a custom PC builder and technician); and even if they do fall into the alienware/dell/acer trap, its usually purchased under advisory from a friend who is enthusiast or a local specialised salesman, or reviews by technically enthusiastic writers on Newegg or amazon. No matter how you slice it, Biases trickle down the experience chain from the top to the bottom. Anyone who knows ANYTHING about PCs knows how often their friends, familly and coworkers ask them about "what computer should I buy" or "is this a good deal".
The prices have nothing to do with the specs of the hardware. Intel sells those chips and the X99 platform for the price it does because they think people will pay that much.
No magic needed. A mobo maker could just use a PLX if they wanted. Expensive, but it works. Most use just 1 video card though.
Some people do that but most don't have PC enthusiasts in the family to build for them or to talk to. Even among family people often buy OEM PC's. After a while it gets tiresome constantly fixing and building your family's PC's. Especially when they blame you for anything that goes wrong since you're the one who built it. There is a reason why many PC enthusiasts walk around with a "No, I will not fix your PC" T-shirt as a half serious gag.
Actually yes it is. The OEM's sell millions of machines yearly for a reason.
KazeoHin
[H]F Junkie
- Joined
- Sep 7, 2011
- Messages
- 9,000
Did you even read my post?
Yup. You're trying to goal post shift to making the argument seem about "people who want a good PC" which is completely different from what you claimed earlier which was that "The problem is that the average, bang-for-buck, Joe buys his PC from an enthusiast." which is obviously false and what I called out as such.
Zarathustra[H]
Extremely [H]
- Joined
- Oct 29, 2000
- Messages
- 38,830
I haven't had a need to buy a motherboard and CPU in over 5 years.
I fully admit to not being up to speed on every last detail of new tech that I haven't researched for my own purposes yet.
I have absolutely zero experience with DDR4. Nothing wrong with that. I don't do the motherboard or RAM reviews. Now if Dan didn't know these things, then there might be a problem.
KazeoHin
[H]F Junkie
- Joined
- Sep 7, 2011
- Messages
- 9,000
Alright, I'll concede that my original wording was false.
My overall point is that most anyone who buys a PC, gets advice from someone who knows better. I sell many, MANY PCs, and I see PCs built by my shop in many small businesses in my area, regardless of that anecdote; 'whitebox' PC sales are a significant fraction of desktop PC sales, but even at bestbuy, newegg, Amazon, the people hired by major OEMs to sell directly to these businesses and the sales prople, are for those businesses are PC enthusiasts to some degree. You can't deny that biases trickle down, especially in a community so dedicated as ours.
There is certainly an effect the enthusiast PC community has on sales, I'm not gonna deny that, but I don't think its a big effect. Certainly not one that will make or break Intel's or AMD's sales.
Plenty of people bought K6's without knowing what CPU was in their OEM PC back in the day and were fine with it. All they knew was that it seemed fast enough and was cheaper than the others. I remember at the time the K6 was largely panned by most enthusiasts due to its relatively ho hum gaming performance too. Only the enthusiasts who were short on cash seemed to like it. Same went for the K5 for that matter. Or any of the Cyrix chips.
A lot of enthusiasts fall into the "fan boy" category and irrationally love particular companies regardless of what's actually better performing or don't understand how to compromise and build for a particular budget.
The laptop buying public likes low res screens that are shiny to make viewing them in anything, but utter darkness impossible.
Those shitboxes you see on the walls for sale in big box stores are actually selling despite being mediocre, expensive, and not particularly upgradeable/expandable. Most people buying computers in stores are buying based on things like which box has the bigger hard drive inside. I haven't heard the "MHz Myth" recently, but ask a lay person what Intel or AMD's model numbers mean and let me know the answer because it's going to be some hilarious nonsense.
Most corporations/institutions are buying X number of boxes at price point Y that have whatever their preferred vendor is selling this year.
Sure, some people do actually ask somebody who knows what they're talking about before buying a PC, but most just find something they can afford with bonus points if it looks cool, is an AIO, or has a touch screen, and go with that and they're fine because a five year old PC with an SSD that's not buried in metric tons of garbage bloatware and malware can run Windows, Office, and surf the Internet just fine.
I'm not going to check Wikipedia to see what the low end graphics card market looks like right now, but are the 64-bit DDR3 cards sporting 6 or 8 GB of VRAM they could never possibly put to use in real time yet?
The laptop buying public likes low res screens that are shiny to make viewing them in anything, but utter darkness impossible.
Those shitboxes you see on the walls for sale in big box stores are actually selling despite being mediocre, expensive, and not particularly upgradeable/expandable. Most people buying computers in stores are buying based on things like which box has the bigger hard drive inside. I haven't heard the "MHz Myth" recently, but ask a lay person what Intel or AMD's model numbers mean and let me know the answer because it's going to be some hilarious nonsense.
Most corporations/institutions are buying X number of boxes at price point Y that have whatever their preferred vendor is selling this year.
Sure, some people do actually ask somebody who knows what they're talking about before buying a PC, but most just find something they can afford with bonus points if it looks cool, is an AIO, or has a touch screen, and go with that and they're fine because a five year old PC with an SSD that's not buried in metric tons of garbage bloatware and malware can run Windows, Office, and surf the Internet just fine.
I'm not going to check Wikipedia to see what the low end graphics card market looks like right now, but are the 64-bit DDR3 cards sporting 6 or 8 GB of VRAM they could never possibly put to use in real time yet?
Singularity_Survivor
Limp Gawd
- Joined
- Apr 7, 2016
- Messages
- 212
D
Deleted member 93354
Guest
Speculating is fun and all, but I say
BRING ON THE FREAKING BENCHMARKS!
Show me the handbrake, lame, 7zip, visual studio 2017 and game benches These are practical applications for enthusiast like us.
I still have apps that take over 2 hours to compile in release mode on xeons.
BRING ON THE FREAKING BENCHMARKS!
Show me the handbrake, lame, 7zip, visual studio 2017 and game benches These are practical applications for enthusiast like us.
I still have apps that take over 2 hours to compile in release mode on xeons.
D
Deleted member 93354
Guest
As a side note I do not that think that amds cross frabic latency issue will be solved anytime soon. That appears to be a long distance between the caches which might be coming up against the maximum speed of electron transfer.
There are ways to mitigate this but they take time. What can be done is create a dirty table of set associative cache between the ccx cores cache that automatically update each other as soon as a slot offset is marked dirty. This would happen as a thread swapped out. But...For concurrent access of similar memory a thread lock on a memory context needs to automatically lock the caches from any updates until we get the all clear signal as to which core has access. (Sort of like two cars fighting for the same parking spot) Latency will still exist here. Data race conditions are a real b1tch for any programmer to deal with. Thankfully Intel and Microsoft created some nice tools to make it a lot nicer for us. But the latency is killer.
There are ways to mitigate this but they take time. What can be done is create a dirty table of set associative cache between the ccx cores cache that automatically update each other as soon as a slot offset is marked dirty. This would happen as a thread swapped out. But...For concurrent access of similar memory a thread lock on a memory context needs to automatically lock the caches from any updates until we get the all clear signal as to which core has access. (Sort of like two cars fighting for the same parking spot) Latency will still exist here. Data race conditions are a real b1tch for any programmer to deal with. Thankfully Intel and Microsoft created some nice tools to make it a lot nicer for us. But the latency is killer.
Tsumi
[H]F Junkie
- Joined
- Mar 18, 2010
- Messages
- 13,752
I would say a small number of enthusiasts actually fall into the fanboy category. The ones that do tend to be extremely loud, and they're the ones you see all the time. Pay attention to who is saying what, and you'll start to see some patterns.
ManofGod
[H]F Junkie
- Joined
- Oct 4, 2007
- Messages
- 12,863
I am suggesting a Ryzen 1600 to a friend for his new build sometime in May or June. Perfect cost, comes with a heatsink already, will be extremely capable for the animation stuff he is doing and going to do and fast! With 32GB of ram, he would be set, especially with a good B350 mainboard.
ManofGod
[H]F Junkie
- Joined
- Oct 4, 2007
- Messages
- 12,863
And everything you just said has absolutely nothing to do with the Ryzen R5 release. Therefore, why don't you start your own thread instead? I am already on a 1700 and 1700X build but, the 1600 with the cooler seems to be the sweat spot for a lot of folks.
This thread has gotten TLDR. Did anyone else read THIS?
http://www.legitreviews.com/one-motherboard-maker-explains-why-amd-am4-boards-are-missing_192470
"Also, their BIOS team and engineers were doing terrible jobs on supporting us on the BIOS microcode updates, driver updates, CPU samples for testing. They have done nothing they should have been doing to support the launch platform partners and always delay or give no response on support requests. We were all having huge issues to debug with limited AMD resource support including validating the parts, and fixing the memory clock speed that is all limited by AMD."
"In general, it’s been too long for AMD to launch a new CPU, so they forgot how to do it, so they launched the CPU just like they were launching the graphics card. They didn’t care about the platform eco-system, so the eco-system is suffering and stock is delayed."
I found that to be really interesting. Basically an "anonymous" source blames AMD for changing the release date too soon and not giving them enough resources the get the boards made and shipped. Interesting.
Apologies if this was covered already
http://www.legitreviews.com/one-motherboard-maker-explains-why-amd-am4-boards-are-missing_192470
"Also, their BIOS team and engineers were doing terrible jobs on supporting us on the BIOS microcode updates, driver updates, CPU samples for testing. They have done nothing they should have been doing to support the launch platform partners and always delay or give no response on support requests. We were all having huge issues to debug with limited AMD resource support including validating the parts, and fixing the memory clock speed that is all limited by AMD."
"In general, it’s been too long for AMD to launch a new CPU, so they forgot how to do it, so they launched the CPU just like they were launching the graphics card. They didn’t care about the platform eco-system, so the eco-system is suffering and stock is delayed."
I found that to be really interesting. Basically an "anonymous" source blames AMD for changing the release date too soon and not giving them enough resources the get the boards made and shipped. Interesting.
Apologies if this was covered already
My post was about the buying habits of the general public, enthusiasts having a minimal impact on it, and how those two things were likely to impact the sales of R5s to the public.
It was in direct to the post before it.
And that post was a direct reply to the twelve posts before it which were also about the R5's sales prospectives.
If you have a problem with that complain to the people who derailed the discussion to the R5's launch for a dozen posts. Or start your own thread about how cool your 1700X is. I don't care really.
Zarathustra[H]
Extremely [H]
- Joined
- Oct 29, 2000
- Messages
- 38,830
When Ryzen 5 was first announced we reported that 6 core models would be two CCX modules with one core each disabled (3+3) and 4 core models would be two CCX modules with two cores each disabled (2+2). There was some concern among our readers based on previous reporting that communication between CCX modules has relatively high latency, and that these layouts might result in lower than ideal performance.
Hardware Unboxed has quantified what type of differences we'll see between 4+0 and 2+2 layouts. From the looks of it, the theory is correct. Splitting things over two CCX modules instead of keeping it all on the same module does have a performance impact, but it seems pretty small to me.
Hardware Unboxed has quantified what type of differences we'll see between 4+0 and 2+2 layouts. From the looks of it, the theory is correct. Splitting things over two CCX modules instead of keeping it all on the same module does have a performance impact, but it seems pretty small to me.
Almost non-existent. OC ability will have a much bigger effect (ie getting 50mhz more on a chip will be worth more than 2+2 vs 4+0).