I'm not a hardware geek, but here's why that option got my vote. AMD is doing great, innovative things and they may well be the wave of the future. However, when it comes to market share, Intel still rules the roost.
no it isn't. AMD have had huge periods where their offerings simply weren't competitive. Now they are doing great but they need to sustain that for a few generations to have any real market effect.
The Democrat policies of universal theft and slavery are as old as mankind. The Republican policies of freedom and property rights are developed from principles only a few hundred years old.
I guess you haven't been reading the news. You're right in that the 18 month lead in process technology they used to have was the true basis for their leadership.
Yes Intel squeezes more transistors into a node than AMD. This means that AMD has to be solid at 7nm before Intel is solid at 10nm in order to outperform Intel.
Guess what. It has happened. The advantage has been overcome. The 18 month lead has been more than overcome. AMD's process deployment now leads Intel, and the forecasts down the road indicate Intel won't be regaining their advantage anytime soon.
As a result, Intel's own CEO has announced a goal (likely stretch) to hold AMD's server CPU, EPYC, to 15-20% of the market (their numbers, not AMDs). That is triple where they are at now and likely a lowball estimate. It was essentially a surprise warning to their investors to try to bake the loss into the stock prices early and reduce the hit when they announce their drops in server CPU revenue.
As much as I dislike Intel and as much as I'd like to cheer them on, Intel is still ahead when it comes to actually shipping silicon and will probably continue to do so for the near future. The only 7nm product AMD plans to start shipping this year is the niche Vega20-based machine learning accelerator which isn't even sampling while Intel has months since gotten to the point of sampling CPUs on their equivalent 10nm process.
The best node AMD has at their disposal right now is the 12nm Globalfoundries on
Intel's 10nm samples appear to be a bit of a parlor trick.
Intel has indicated that the multi-patterning process used to produce them is producing too many yield defects to produce 10nm cost-effectively. In the past, when a statement like that was made it meant that a bit more tweaking / learning would get them there. This time, they are saying "we understand the yield issues and have defined improvements for them, but they will take time to implement and qualify" and they are not committing to volume produc
Nice cherrypick there. Now that multiple cores have been mainstream for over a decade, the single-core argument has lost steam. Software has adjusted. Ryzen beats the 8700K in most multicore and real-world application workloads and wins soundly in dollars / performance. The comedy in the single-core benchmark is highlighted by the fact that the Pentiums still outrank many of the most modern processors.
In addition, this single core performance win on Intel's part is related to their aggressive approach to sp
According to those benchmarks, a 16-core Threadripper is faster than a 32-core EPYC. It doesn't look very trustworthy to me. Given past precedents, I would take any closed-source benchmark result with a skyscrapper-sized grain of salt.
And a lot more complex than clock speeds. The days of single-thread importance are gone. But, there next process step is able to hit 5GHz if they choose to optimize in that direction. They won't because that isn't what's important.
I bought processors from both companies in the past. Probably more from Intel, but mostly because Intel happened to rule the market when I had more funds to spend on systems. AMD typically only seems to make a dent in Intel's sales when Intel makes a series of missteps, miscalculations, and allows slothful hubris to take over.
Last time, Intel pushed for more power, Netburst, RD RAM, and Itanium focusing on the Gigahertz War. AMD actually looked at x64 instructions, giving people a way to migrate applicatio
Remember when the Athlon first came out? Faster than anything Intel had. How about the x86-64 architecture? Linux was running 64 bit on AMD days after x86-64 was released. We are still running many of our 64 core AMD Opterons. For some tasks 64 general purpose computing units in one machine are hard to beat. And now Ryzen chips are kicking Intel butt too.
People who say "finally" have very short memories and attention spans.
Remember when the Athlon first came out? Faster than anything Intel had.
I do remember, and I remember having a string of PCs (over a span of 6-7 years) powered exclusively by AMD processors.
However at some point AMD lost its edge, Intel reigned supreme and it seemed that AMD only survived on the back of its acquisition of ATI. It's great that AMD is back in the game seriously. My next PC might just be AMD powered again.
Intel need to have their asses kicked from time to time to get to the next architecture of chips. If we didn't had an MOS/ZiLOG, a Motorola, a IBM and a AMD, the intel chips would still be 8080's running at 100Mhz.
I have a raspberry pi doing something. I am quite impressed with it and it does no have the many processor flaws in amd or intel.
It already runs on linux. This is not to say arm does not have issues. I would think seriously about running arm architecture on a real motherboard. if it came to be an option.
The problem with the Pi is that it is rather anemic (which is understandable given its audience and price point). A Pi is never going to be able to replace your desktop. The most glaring limitation is only 1GB of RAM, combined with a lack of SATA for fast storage.
As far as ARM goes in general, I don't see ARM replacing x86 any time soon, simply due to market inertia. A lot of people need desktop apps that just don't run on ARM.
As sjwest points out, ARM is probably the future if the strategy is right. x86[-64] is an awful architecture full of backward compatibility kludges and complexity. ARMv8 is clean, efficient and well understood.
What is needed for this to happen is modularity, i.e. decouple the GPU from the CPU and allow discrete components on a PCIe bus. Similar with memory. Slot it, don't solder the bugger to the board. Some of this is already happening but the ARM ecosystem still focuses mainly on SoC models, which is not what we general purpose computing bods want. If I could have an octa-core A53 with slotted DDR3, Gig-E on PCIe, USB3, SATA/NVME and a discrete graphics processor and, more importantly, the choice of these components rather than someone else's SoC design, I'd drop x86 like a hot rock.
I agree with all of this, but I'd also be willing to put a few bob on Intel dominating the future with an implementation of RISC-V.
Competition seems to have been a good thing for Intel:
—AMD added 64-bit to x86, forcing Intel to match it and ultimately abandon Itanic.
—ARM has gradually educated the world to adopt a new, cleaner architecture in a way that the major RISC players of the 90s (IBM Power, Sun SPARC, H-P PA-RISC, DEC Alpha, SGI and others with MIPS) never managed. This has undoubtedly influenced Intel for the better, apart from their messy x86 ISA that we've been expecting to die soon for the last 30 years.
Perhaps the world is finally ready to be weaned-off the x86 legacy for good. ARM may have too many IP/licensing issues for Intel, whereas there's no such issue for RISC-V, and Intel announced an interesting investment last month: https://riscv.org/2018/05/sifi... [riscv.org]
But whatever happens, we need the likes of AMD to keep Intel on their toes.
From what I know about Risc-V, it certainly looks interesting, but it will have to overcome the same hurdles an any other new architecture. When Apple moved from Motorola to Power PC, they went through a few difficult years before they had really convincing products (and software!) again.
I guess Intel could survive that, at the expense of losing quite a bit of market share for a while. Much like AMD survived their years of being behind with Bulldozer. But are they willing to take the risk? Large corporation
Later generations of RISC instruction set - PowerPC, ARM and Alpha - don't have any of this nonsense. There are no delayed branches, no rotating register files, and as an assembly programmer, the only reason to understand the details of the pipeline implementation would be to maximise performance. These instruction sets are much nicer to use. Design err
Some of the biggest advances in CPU design involved incorporating more functionality onto the CPU chip: DRAM controller, cache, vector processor, floating point, memory mapping, multi-core. Taking the GPU off-die is a step backwards, and introduces extra delays for processes where the GPU is used as a math processor rather than a display driver.
If I ever got a mod point I might give your comment an insightful mod. Unclear if you know why, or just stumbled on it by accident.
The CPUs have maxed out as far as the human perspective is concerned. The software is still a huge problem, but the hardware is already far beyond our ken and we can't perceive the differences anymore. "I don't care" is the only reasonable assessment of imperceptible differences between CPUs. I just bought a new wrist toy that has more computing power than I can assess.
Oh and besides: they're companies. They can fail or survive for a myriad of reasons, of which the quality of their product is only a small part (I'm looking at you, Microsoft and FaceBook).
So is it time to talk about solutions? My own favored solution approach would be a progressive tax on corporate profits based on market share. The goal is to increase freedom by encouraging the real competition that offers choices while driving innovation.
As it would apply in the case of Intel, the corporation's profits would almost surely be taxed at the highest tax rate. Some of the tax money would be used to monitor and even regulate Intel to make sure they aren't abusing their strong position in the mar
General Electric has been following your policy for decades, spinning off divisions that were less profitable or not part of their core mission. The result? GE has been removed from the Dow Jones Industrial Index.
NO, that IS NOT my suggestion. Go back and read it again.
If you don't understand it, but want to, then you should ask a question.
If you actually understand it and have a suggestion to improve it, or even a constructive criticism based on actual understanding, then that would be welcome.
If you have nothing to say, then you should say nothing.
Yep.
I was looking at my computer and the last time I upgraded the CPU was six years ago. I got a new video card 3 years ago.
I play lots of games and do lots of virtualization and have no issues with the highest or 2nd highest graphics settings for games and running the vm I need.
The CPU and upgrade wars are long over and no one cares.
Lots of people still care. This osn't so much about graphics but about being able to do lots of calculations really fast, for 3D rendering, bitcoin mining, simulating supernovae etc.
Next year AMD will have 7nm server products superior in every way to Intel's offerings and it's going to remain that way for an entire year, maybe more. You have a 48 core 7nm Zen2 product versus 28 core "Cooper lake", which is just a warmed up Skylake SP and this is because Intel's 10nm process has failed so completely. It'll be 2006 all over again, when Opterons had 25% of the market.
I give AMD for a least matching Intel with Ryzen and Ryzen 2. But and its a big but, AMD has yet to convince PC makers that it is preferred over Intel with buyers. What could win over more Intel users to AMD CPU's is more exploits against Intel chips and AMD remaining mostly immune. For me the Spectre Meltdown weaknesses have not concerned me as much as the lousy performance robbing fixes that were implemented.
I am under the impression in the server space, some of the Epyc gear is fantastic, due to all the PCI-e lanes and cores. On the desktop, Ryzen is ok, not amazing. A lot better than their old stuff.
It's nice to see them catch most of the way back up though, considering Intel has been taking a beating (10nm CPUs are baffling Intel right now) we might see the next gen AMD CPUs in 18 months actually compete entirely (single core performance, multicore, power use, heat, etc)
I mean, sure Intel is starting to circle the drain. Its semi-conductor manufacturing has gone from a leading the industry to being behind, and its only diversification has so far been into far smaller markets than its primary one of making CPUs. It also doesn't appear to have any ace up its sleeve about post silicon chips, SI has less than a decade left of getting any better at all, and the improvements it'll make in that time aren't even close to the ones it used to make.
But that doesn't mean AMD is the one that's going to defeat it. There's a highly cost effective ARM server out there now with the Thunder X2. And with the vast potential of post silicon materials possibly being thousands of times faster or more (in clock speed) and using far less power whoever gets to commercialize the "right" solution first could have a nigh unstoppable advantage. And since there's no definitive answer as to what comes after SI yet, Intel doesn't have any huge advantage in finding and commercializing it, not that they seem that concentrated on even doing so.
Desktop processors seem to be waning into a commodity item. Progress is slow. Hard to get excited about bigger/better/faster processors for my home PC that only gets fired up weekly or so.
As computing shifts more and more towards mobile, x86 will, over time, become less relevant. As mobile processors (e.g. ARM) become more advanced, they are likely to make inroads into spaces currently ruled by x86 (this has already begun with some small laptops). With more money going into non-x86 processors, it will not be too long before x86 is completely and utterly outclassed.
AMD's Ryzen and Threadripper designs are really impressive, though, and they're likely to pick up some marketshare for a little
If I were to blow lots of money on an 8 core system with ultra expensive DDR 4 ram since ram has went up 200% I can get an Intel Skylake X for just $200 with a more modern architecture.
AMD has been first to market by a wide margin in a few categories, before (consumer grade 64 bit processors come to mind, for one), and have had indisputably the faster processors in generations where they were not first. Every time they have a clear advantage over Intel, they somehow trip over their own feet and just hand the entire market back to Intel. Really keen to see how they fuck up a solid, possibly year long, lead.
I'm only concerned with whatever is going into my next Alienware laptop when it's time to replace the old one, and that it can keep up with the latest games. =P
The thing every post I've read seems to miss is what AMD has done to undercut Intel on price. The infinite fabric, yes lame marketing term, is why Ryzen/Threadripper are the price to performance leader. Chips that for Intel would be a 4 core i5 at best are being strapped together as to make an 8 core or 16 core for AMD. Even a bad fab with 2 usable cores is usable, where Intel might as well scrap it. This is upping their yields greatly and why Intel can't compete on price. A 16 core Intel has is fabbed as a
I bought a laptop a couple of years back with an AMD Radeon HD 6490M integrated mobile graphics processor, and AMD refused to release working drivers for it. ONE driver version was released upon launch. Downloading and installing any updated drivers from AMD always failed.
When contacting them about this, their customer support totally ignored it.
I will never ever buy any AMD product again and recommend everyone else to stay away from AMD.
AMD release broken garbage and refuses to support their paying cus
One of the core commandments of CPU throughput is still valid -- Know Thy Workload. AMD has been disruptive in applications that can take advantage of multithreading and lots more cores and PCIe lanes. Expect Intel to counter though (and they already have started). For parallel workloads, Threadripper / Ryzen have a great bang for your buck and the next iteration of that tech will likely be even better.
However -- for some workloads, single threaded performance still rules, and Intel is the obvious winner
I am building a new data management system that can do lots of operations (e.g. file system, database, NoSQL, etc.) in parallel. The more threads I have, the faster individual queries execute. A single query on a 100 million row DB table with 25 columns will currently execute about 50% faster on my hex-core than it does on my quad core. I can't wait to benchmark it on one of those ThreadRipper 2 machines (32 core, 64 threads) when they start shipping later this year.
Obviously, enough cache to keep the whole table loaded into memory is the best option. If each row in the table has about 250 bytes in it, you can still cache the whole thing using 25 GB of RAM. Buying a machine these days with 32 GB does not break the bank, even with RAM prices higher than they were a couple years ago.
Since this system implements relational tables as a columnar store, it only needs to read in the columns that are part of the query. If you have the 25 columns in my example, if you run a q
Hotels are tired of getting ripped off. I checked into a hotel and they
had towels from my house.
-- Mark Guido
Poorly designed poll (Score:4)
One poll option incorporates two others, and surprise, surprise, it's by far the most popular.
Re: (Score:2)
Re: (Score:2)
Re: (Score:1)
After seeing Trump I would argue its a radical difference than Obama
Re: (Score:1)
Re: (Score:3)
Yes and No. The poll is doing a great job, but your comment makes sense.
run? (Score:3)
AMD's very presence has kept intel on their toes for years now. intel's only advantage is size; their chips aren't particularly special.
Re:run? (Score:5, Interesting)
I guess you haven't been reading the news. You're right in that the 18 month lead in process technology they used to have was the true basis for their leadership.
Yes Intel squeezes more transistors into a node than AMD. This means that AMD has to be solid at 7nm before Intel is solid at 10nm in order to outperform Intel.
Guess what. It has happened. The advantage has been overcome. The 18 month lead has been more than overcome. AMD's process deployment now leads Intel, and the forecasts down the road indicate Intel won't be regaining their advantage anytime soon.
As a result, Intel's own CEO has announced a goal (likely stretch) to hold AMD's server CPU, EPYC, to 15-20% of the market (their numbers, not AMDs). That is triple where they are at now and likely a lowball estimate. It was essentially a surprise warning to their investors to try to bake the loss into the stock prices early and reduce the hit when they announce their drops in server CPU revenue.
Re: (Score:1)
AMD had basically no share of the server market before.
So I guess if the intention is to support the weaker player to get a more competitive market then the companies aren't playing it.
7 nm is supposed to have twice the density whereas Intel claim 2.7x for 10 nm ... The sizes I've seen before haven't implied that but what do I know.
One unlikely should watch them as 16, 14, 12, 10 and 7 and compare staight up against each other.
So in density Intel may be fine even with their 10 nm vs GlobalFoundries 7 nm but
Re: run? (Score:1)
AMD Opteron gained a fair chunk of market share back in the day, you cant say they never had that share.
Re: (Score:1)
And I didn't.
I mean in more near-term history. They were down to ~0% market share.
Re: (Score:2)
AMD's process deployment now leads Intel
As much as I dislike Intel and as much as I'd like to cheer them on, Intel is still ahead when it comes to actually shipping silicon and will probably continue to do so for the near future. The only 7nm product AMD plans to start shipping this year is the niche Vega20-based machine learning accelerator which isn't even sampling while Intel has months since gotten to the point of sampling CPUs on their equivalent 10nm process.
The best node AMD has at their disposal right now is the 12nm Globalfoundries on
Re: (Score:2)
Intel's 10nm samples appear to be a bit of a parlor trick.
Intel has indicated that the multi-patterning process used to produce them is producing too many yield defects to produce 10nm cost-effectively. In the past, when a statement like that was made it meant that a bit more tweaking / learning would get them there. This time, they are saying "we understand the yield issues and have defined improvements for them, but they will take time to implement and qualify" and they are not committing to volume produc
Re: (Score:2)
The advantage has been overcome.
Except... [cpubenchmark.net]
Re: (Score:2)
Nice cherrypick there. Now that multiple cores have been mainstream for over a decade, the single-core argument has lost steam. Software has adjusted. Ryzen beats the 8700K in most multicore and real-world application workloads and wins soundly in dollars / performance. The comedy in the single-core benchmark is highlighted by the fact that the Pentiums still outrank many of the most modern processors.
In addition, this single core performance win on Intel's part is related to their aggressive approach to sp
Re: (Score:1)
Re: (Score:2)
Re: (Score:2)
I bought processors from both companies in the past. Probably more from Intel, but mostly because Intel happened to rule the market when I had more funds to spend on systems.
AMD typically only seems to make a dent in Intel's sales when Intel makes a series of missteps, miscalculations, and allows slothful hubris to take over.
Last time, Intel pushed for more power, Netburst, RD RAM, and Itanium focusing on the Gigahertz War. AMD actually looked at x64 instructions, giving people a way to migrate applicatio
AMD has been kicking butt for years, kids (Score:2)
Remember when the Athlon first came out? Faster than anything Intel had.
How about the x86-64 architecture? Linux was running 64 bit on AMD days after x86-64 was released.
We are still running many of our 64 core AMD Opterons. For some tasks 64 general purpose computing units in one machine are hard to beat.
And now Ryzen chips are kicking Intel butt too.
People who say "finally" have very short memories and attention spans.
Re:AMD has been kicking butt for years, kids (Score:4, Informative)
Remember when the Athlon first came out? Faster than anything Intel had.
I do remember, and I remember having a string of PCs (over a span of 6-7 years) powered exclusively by AMD processors.
However at some point AMD lost its edge, Intel reigned supreme and it seemed that AMD only survived on the back of its acquisition of ATI. It's great that AMD is back in the game seriously. My next PC might just be AMD powered again.
Re: (Score:2)
Intel need to have their asses kicked from time to time to get to the next architecture of chips.
If we didn't had an MOS/ZiLOG, a Motorola, a IBM and a AMD, the intel chips would still be 8080's running at 100Mhz.
Other... (Score:2, Funny)
Cyrix will rise again!
Re: (Score:2)
Re: Other... (Score:1)
Transmeta rulez.
I still like Motorola chips, too.
What am I, psychic? (Score:2)
Is this a rhetorical poll?
No Cowboy Neal option? (Score:2)
Cowboy Neal is in your CPU spying on you!
Re: (Score:2)
i thought he left Intel?
Re: (Score:2)
Ha!
Another option: Cowboy Neal Inside (TM)
Re: (Score:2)
right about now, Intel will jump on any ideas :)
Intel or AMD? Maybe neither. (Score:3, Insightful)
Both are threatened by ARM cores in all sorts of digital products that are not traditional PCs, that are displacing PCs.
This goes back and forth constantly (Score:2)
arm chips ? (Score:3)
I have a raspberry pi doing something. I am quite impressed with it and it does no have the many processor flaws in amd or intel.
It already runs on linux. This is not to say arm does not have issues. I would think seriously about running arm architecture on a real motherboard. if it came to be an option.
Re: (Score:3)
The problem with the Pi is that it is rather anemic (which is understandable given its audience and price point). A Pi is never going to be able to replace your desktop. The most glaring limitation is only 1GB of RAM, combined with a lack of SATA for fast storage.
As far as ARM goes in general, I don't see ARM replacing x86 any time soon, simply due to market inertia. A lot of people need desktop apps that just don't run on ARM.
Re: (Score:2)
Not a desktop but Win10 arm 2 in 1 is already on the market.
https://www.asus.com/us/2-in-1... [asus.com]
https://store.hp.com/us/en/pdp... [hp.com]
https://www3.lenovo.com/us/en/... [lenovo.com]
ARM (Score:3)
What is needed for this to happen is modularity, i.e. decouple the GPU from the CPU and allow discrete components on a PCIe bus. Similar with memory. Slot it, don't solder the bugger to the board. Some of this is already happening but the ARM ecosystem still focuses mainly on SoC models, which is not what we general purpose computing bods want. If I could have an octa-core A53 with slotted DDR3, Gig-E on PCIe, USB3, SATA/NVME and a discrete graphics processor and, more importantly, the choice of these components rather than someone else's SoC design, I'd drop x86 like a hot rock.
Re:ARM —or RISC-V? (Score:5, Insightful)
I agree with all of this, but I'd also be willing to put a few bob on Intel dominating the future with an implementation of RISC-V.
Competition seems to have been a good thing for Intel:
—AMD added 64-bit to x86, forcing Intel to match it and ultimately abandon Itanic.
—ARM has gradually educated the world to adopt a new, cleaner architecture in a way that the major RISC players of the 90s (IBM Power, Sun SPARC, H-P PA-RISC, DEC Alpha, SGI and others with MIPS) never managed. This has undoubtedly influenced Intel for the better, apart from their messy x86 ISA that we've been expecting to die soon for the last 30 years.
Perhaps the world is finally ready to be weaned-off the x86 legacy for good. ARM may have too many IP/licensing issues for Intel, whereas there's no such issue for RISC-V, and Intel announced an interesting investment last month:
https://riscv.org/2018/05/sifi... [riscv.org]
But whatever happens, we need the likes of AMD to keep Intel on their toes.
Re: (Score:1)
From what I know about Risc-V, it certainly looks interesting, but it will have to overcome the same hurdles an any other new architecture. When Apple moved from Motorola to Power PC, they went through a few difficult years before they had really convincing products (and software!) again.
I guess Intel could survive that, at the expense of losing quite a bit of market share for a while. Much like AMD survived their years of being behind with Bulldozer. But are they willing to take the risk? Large corporation
ARM "educated" a "cleaner" architecture? (Score:3)
ARM did not build a "clean" design from the perspective of a branch predictor.
https://www.jwhitham.org//2016/02/risc-instruction-sets-i-have-known-and.html [jwhitham.org]
Re: (Score:3)
ARMv8 is clean, efficient and well understood.
OK, but if it doesn't run x86 code you might as well forget it.
Re: (Score:2)
Other (Score:4)
I voted Other because I don't care.
Re: (Score:2)
I voted Other because I don't care.
If I ever got a mod point I might give your comment an insightful mod. Unclear if you know why, or just stumbled on it by accident.
The CPUs have maxed out as far as the human perspective is concerned. The software is still a huge problem, but the hardware is already far beyond our ken and we can't perceive the differences anymore. "I don't care" is the only reasonable assessment of imperceptible differences between CPUs. I just bought a new wrist toy that has more computing power than I can assess.
Different
Re: (Score:2)
That is indeed just what I meant. Any CPU or GPU is good enough for my needs.
Re: (Score:2)
Oh and besides: they're companies. They can fail or survive for a myriad of reasons, of which the quality of their product is only a small part (I'm looking at you, Microsoft and FaceBook).
Time for solutions? (Score:3)
So is it time to talk about solutions? My own favored solution approach would be a progressive tax on corporate profits based on market share. The goal is to increase freedom by encouraging the real competition that offers choices while driving innovation.
As it would apply in the case of Intel, the corporation's profits would almost surely be taxed at the highest tax rate. Some of the tax money would be used to monitor and even regulate Intel to make sure they aren't abusing their strong position in the mar
Re: (Score:2)
Re: (Score:2)
NO, that IS NOT my suggestion. Go back and read it again.
If you don't understand it, but want to, then you should ask a question.
If you actually understand it and have a suggestion to improve it, or even a constructive criticism based on actual understanding, then that would be welcome.
If you have nothing to say, then you should say nothing.
Re: (Score:2)
I voted Other because I don't understand.
Re: (Score:1)
I would mod you up, but I don't care.
Re: (Score:2)
I was looking at my computer and the last time I upgraded the CPU was six years ago. I got a new video card 3 years ago.
I play lots of games and do lots of virtualization and have no issues with the highest or 2nd highest graphics settings for games and running the vm I need.
The CPU and upgrade wars are long over and no one cares.
Re: (Score:2)
Lots of people still care. This osn't so much about graphics but about being able to do lots of calculations really fast, for 3D rendering, bitcoin mining, simulating supernovae etc.
Next year is AMD's year (Score:1)
Next year AMD will have 7nm server products superior in every way to Intel's offerings and it's going to remain that way for an entire year, maybe more. You have a 48 core 7nm Zen2 product versus 28 core "Cooper lake", which is just a warmed up Skylake SP and this is because Intel's 10nm process has failed so completely. It'll be 2006 all over again, when Opterons had 25% of the market.
Re: (Score:2, Funny)
You heard it here first: 2019 will be the year of Linux on the AMD Desktop!
Re: (Score:2)
Are you unaware of the hundreds of free games, many of them excellent, developed with the Tomb Raider Level Editor?
AMD has momentum, Intel still rules (Score:1)
I give AMD for a least matching Intel with Ryzen and Ryzen 2. But and its a big but, AMD has yet to convince PC makers that it is preferred over Intel with buyers. What could win over more Intel users to AMD CPU's is more exploits against Intel chips and AMD remaining mostly immune. For me the Spectre Meltdown weaknesses have not concerned me as much as the lousy performance robbing fixes that were implemented.
They're getting closer (Score:2)
I am under the impression in the server space, some of the Epyc gear is fantastic, due to all the PCI-e lanes and cores. On the desktop, Ryzen is ok, not amazing. A lot better than their old stuff.
It's nice to see them catch most of the way back up though, considering Intel has been taking a beating (10nm CPUs are baffling Intel right now) we might see the next gen AMD CPUs in 18 months actually compete entirely (single core performance, multicore, power use, heat, etc)
As a NAS enthusiast, I'm frustrat
Other, AMD finally giving Intel run for its money (Score:2)
Intel down the drain (Score:3)
But that doesn't mean AMD is the one that's going to defeat it. There's a highly cost effective ARM server out there now with the Thunder X2. And with the vast potential of post silicon materials possibly being thousands of times faster or more (in clock speed) and using far less power whoever gets to commercialize the "right" solution first could have a nigh unstoppable advantage. And since there's no definitive answer as to what comes after SI yet, Intel doesn't have any huge advantage in finding and commercializing it, not that they seem that concentrated on even doing so.
Re: (Score:2)
[citation needed]
In addition, there's no material that allows signal propagation faster than 30 cm/ns.
Best buggy whips (Score:2)
Desktop processors seem to be waning into a commodity item. Progress is slow. Hard to get excited about bigger/better/faster processors for my home PC that only gets fired up weekly or so.
Other: x86 is (slowly) dying (Score:2)
As computing shifts more and more towards mobile, x86 will, over time, become less relevant. As mobile processors (e.g. ARM) become more advanced, they are likely to make inroads into spaces currently ruled by x86 (this has already begun with some small laptops). With more money going into non-x86 processors, it will not be too long before x86 is completely and utterly outclassed.
AMD's Ryzen and Threadripper designs are really impressive, though, and they're likely to pick up some marketshare for a little
Great a 2014 era 4770K with 8 cores (Score:2)
It is 5 years behind. Citation is here. The 4790K is still 5% faster in single core. AMD is bad for games but ok if you compile lots of code or run many virtual machines. [userbenchmark.com]
If I were to blow lots of money on an 8 core system with ultra expensive DDR 4 ram since ram has went up 200% I can get an Intel Skylake X for just $200 with a more modern architecture.
Eager to see how AMD will fuck this up (Score:2)
AMD has been first to market by a wide margin in a few categories, before (consumer grade 64 bit processors come to mind, for one), and have had indisputably the faster processors in generations where they were not first. Every time they have a clear advantage over Intel, they somehow trip over their own feet and just hand the entire market back to Intel. Really keen to see how they fuck up a solid, possibly year long, lead.
AMD / INTEL will both lose (Score:1)
Re: (Score:2)
With the mobile market increasing and the desktop/laptop market crawling, ARM will beat them both out.
And Apple will then be at the top of the heap...
Who cares? (Score:1)
AMD's current advantage (Score:1)
AMD is awful (Score:1)
Single versus multi threaded (Score:2)
One of the core commandments of CPU throughput is still valid -- Know Thy Workload. AMD has been disruptive in applications that can take advantage of multithreading and lots more cores and PCIe lanes. Expect Intel to counter though (and they already have started). For parallel workloads, Threadripper / Ryzen have a great bang for your buck and the next iteration of that tech will likely be even better.
However -- for some workloads, single threaded performance still rules, and Intel is the obvious winner
More threads the better (Score:2)
Re: (Score:2)