AMD Unveils the World's Most Powerful Desktop CPUs (zdnet.com) 187
ZDNet reports:
In the never ending war between the chip giants, AMD has released a salvo by unveiling what are the world's most powerful desktop processors -- the new 24-core AMD Ryzen Threadripper 3960X and 32-core AMD Ryzen Threadripper 3970X... These 3rd-generation Ryzen Threadripper Processors are built using AMD's 7-nanometer "Zen 2" core architecture, and both chips feature 88 PCIe 4.0 lanes with extraordinary power efficiency.
On the performanced front, AMD claims that the new 32-core Ryzen Threadripper 3970X offers up to 90 percent faster performance over the competition... This performance doesn't mean the chips are power-hungry either, with AMD claiming they deliver up to 66 percent better power efficiency compared to previous generation processors. The new chips do, however, need a new socket. The new socket is called sTRX4, which offers expansion for serious multi-GPU and NVMe arrays, quad channel DDR4, ECC support, and unlocked overclocking.... [T]hey both will be available starting Tuesday, November 19.
Engadget reports: After getting some wins against Intel in the desktop enthusiast processor race, AMD is trying to run up the score with its latest model, the Ryzen 9 3950X. It has 16 cores/32 threads, a 3.5 Ghz base clock with up to 4.7 GHz boost (on two cores) and 105 watt power consumption (TDP), and costs $749, compared to $1,199 for Intel's 12-core i9-9920X. At the same time, AMD claims it outperforms the i9-9920X in gaming and even more so for content creation, where those extra cores can be best exploited.
According to the company, it'll do some Adobe Premiere tasks up to 26 percent quicker than an i9-9920X, and 42 percent faster than an 8-core i9-9900K. Better still, the Ryzen 9 3950X delivers 2.34 times more performance per watt than its Intel counterpart, and consumes 173W of absolute wall power compared to 304W for the i9-9920X. The power figures alone could be decisive for creators who run multiple workstations for 3D animation and rendering...
If $749 is $700 too much, AMD has another option -- the Athlon 3000G. The dual-core processor runs at 3.5Ghz, but AMD said it's "the only unlocked option in its segment," meaning you can push it to around 3.9Ghz. That'll boost its performance ahead of Intel's $73 Pentium G5400, AMD said. The Athlon 3000G will arrive November 19th for $49.
On the performanced front, AMD claims that the new 32-core Ryzen Threadripper 3970X offers up to 90 percent faster performance over the competition... This performance doesn't mean the chips are power-hungry either, with AMD claiming they deliver up to 66 percent better power efficiency compared to previous generation processors. The new chips do, however, need a new socket. The new socket is called sTRX4, which offers expansion for serious multi-GPU and NVMe arrays, quad channel DDR4, ECC support, and unlocked overclocking.... [T]hey both will be available starting Tuesday, November 19.
Engadget reports: After getting some wins against Intel in the desktop enthusiast processor race, AMD is trying to run up the score with its latest model, the Ryzen 9 3950X. It has 16 cores/32 threads, a 3.5 Ghz base clock with up to 4.7 GHz boost (on two cores) and 105 watt power consumption (TDP), and costs $749, compared to $1,199 for Intel's 12-core i9-9920X. At the same time, AMD claims it outperforms the i9-9920X in gaming and even more so for content creation, where those extra cores can be best exploited.
According to the company, it'll do some Adobe Premiere tasks up to 26 percent quicker than an i9-9920X, and 42 percent faster than an 8-core i9-9900K. Better still, the Ryzen 9 3950X delivers 2.34 times more performance per watt than its Intel counterpart, and consumes 173W of absolute wall power compared to 304W for the i9-9920X. The power figures alone could be decisive for creators who run multiple workstations for 3D animation and rendering...
If $749 is $700 too much, AMD has another option -- the Athlon 3000G. The dual-core processor runs at 3.5Ghz, but AMD said it's "the only unlocked option in its segment," meaning you can push it to around 3.9Ghz. That'll boost its performance ahead of Intel's $73 Pentium G5400, AMD said. The Athlon 3000G will arrive November 19th for $49.
Longevity (Score:3, Interesting)
Ryzen looks like a great option for use today, but it's a bit thin on PCIe lanes. You get basically 16x for a GPU and 4x for everything else, which could be entirely taken up by a single NVMe SSD. Then all your additional lanes are on the chipset and share a 4x link to the CPU/RAM.
That might be a bit limiting in the long term. We already have USB 3.2 and Thunderbolt that will saturate the chipset's bandwidth. You can split the 16x link into to 8x slots with minimal performance loss but still... Expect motherboards to have a bunch of 1x slots and a compromised second NVMe slot that splits lanes from the first one.
That may not matter, it might be better to save money now over Threadripper and just upgrade in a few years when it becomes an issue. Also Threadripper tends not to perform so well in games so if you game it might not be an option.
Re:Longevity (Score:5, Informative)
The news is about the new generation of Threadripper; with the currently available TRX40 chipset advertized by AMD as "72 available PCIe 4.0 lanes" [heise.de]
You get 56 lanes directly from the processor. And 16 from the chipset. The connection between processor and chipset is 8 lanes, which would only be a bottle neck if you really use PCIe 4.0 speed on the 16 lanes from the chipset.
If you need more than 72 lanes, you'll have to wait for boards with the TRX80 or WRX80 chipset.
Re: (Score:2)
Threadripper is more than enough for almost any workstation type scenario. Since I don't play modern games much (well, maybe FS 2020 when it comes out) it's tempting to go for that and future proof the system in the expectation that I'd keep it for many years.
My current system has parts dating back to 2007, with the CPU and mobo from 2012. It's the lack of PCIe lanes that is pushing me to upgrade it, nothing else.
Re:Longevity (Score:5, Informative)
In the example given in TFS, they are comparing it to i9-9920X which has 44 lanes, but they are PCI-E 3.0.
AMD is only at slight disadvantage while the GPU, Storage etc. manufacturers catch up and switch over to PCI-E 4.0. If you use 3.0/3.1 devices in 4.0 slots you will be essentially wasting half of the bandwidth of those slots. Kinda like plugging USB2 devices in USB3 slots.
Re: (Score:3)
Thing is many cards are only PCIe 3.0 and since Intel is not going to support 4.0 at all (they are going straight to 5.0 apparently) maybe there won't be that many.
At this point it's really hard to know if PCIe 4.0 is helpful or not.
Re: (Score:2)
Re: (Score:2)
if Intel is going straight to PCIe 5.0 then only because they were so lazy and overslept PCIe 4.
Or they were preoccupied with their yield issues, the latter seeming more probable.
Re: (Score:2)
At this point it's really hard to know if PCIe 4.0 is helpful or not.
It's helpful for the chipset and the NVMe which is where you're likely to see a limitation if you have some strange use case that actually is able to push the PCIe lanes to their limit. Currently GPUs don't even make full utilization of the PCIe 3.0 lanes they have.
Re:Longevity (Score:4, Informative)
the new Ryzen desktop chips have 16x for GPU, 4x dedicated for an m.2, and then another 4xx dedicated for other stuff. These are also PCIE4 lanes, so this is equivelent to an Intel chip having 32x for GPU, 8x dedicated for m.2, and 8x dedicated for other stuff. It is only the rare case where you are going to overflow this with a desktop system, and in that case, you are now in HEDT territory, which is what Threadripper is for.
Re: (Score:3)
4x for everything else, which could be entirely taken up by a single NVMe SSD
The new X570 chipset includes PCIe 4.0 support which means x4 would be almost 8GB/s (64Gb/s) throughput. I don't think there's any concerns of saturating that with a single NVMe SSD.
Re: (Score:2)
Corsair Force MP600 has read speeds just shy of 5GB/sec, so if you had two they would be constrained by lack of PCIe lanes. People make RAID0 stripes out of them now with 9GB/sec+ performance, and of course they will continue to get faster over time.
Maybe 9GB/sec isn't something you need... But I'm expecting there to be some interesting applications in future. This is as fast as DDR3 RAM so although there is additional latency compared to RAM I think it's hard to predict where this will go in future.
Re: (Score:3)
Re: (Score:2)
but it's a bit thin on PCIe lanes
Ryzen 3 provides 16x PCIe 4.0 lanes to the GPU, 4x PCIe 4.0 lanes to the NVMe slot, and 4x PCIe 4.0 lanes to the chipset.
Nothing on the consumer market saturates those links. This will become relevant if you're running a highspeed renderbox, or throwing multiple GPUs in for an AI calculation farm.
It's not a case of a few years either. We're in a realm where GPUs don't even saturate the 16x PCIe 3.0 lanes, and they have double the bandwidth to play with now. NVMe storage is already blazingly fast to the poin
Re: (Score:2)
Re: (Score:2)
Intel has ever fewer PCIe lanes on their consumer models and their Xeon stuff is way over-priced compared to AMD.
Re: Longevity (Score:2)
"Optimized for liquid cooling" (Score:5, Insightful)
I saw "Optimized for liquid cooling" on the marketing blurb. That's an entertainingly optimistic way of saying "too hot for air cooling".
Re:"Optimized for liquid cooling" (Score:5, Interesting)
No what it means is that the form factor design for an air cooler would be prohibitive to the current design of minimalist cases. And minimialist designs have been the big thing in the enthusist market for the last 7-8 years. Look at say the cooler master hyper 212 and variations, they're still rated as one of the best aftermarket coolers to have ever been made. They're quiet, have good air flow, but they're big, heavy, and have serious installation issues. Everything from causing motherboards to break to absolute pain in the ass installation.
Liquid cooling on the other hand, especially closed loop designs have come so far down in price that they're on par with the cost of the 212 and other quiet but beefy designs.
Re:Liquid cooling is a PITA too though. (Score:5, Informative)
1: There are chipset blocks to keep the VRMs and other important mobo chips cool
2: The sound of 2 pumps and 6-12 large fans turning slowly (or not at all) is infinitely more quiet than screaming GPU and CPU fans
3: Efficiently removing the heat from your components dramatically increases their lifespan
Over the 2 builds / 10 years I have been running high end watercooled home gaming boxes, never had a leak or component failure (apart from a HDD) even if clocking the living shit out of them.
Re: (Score:3)
Re: (Score:3)
For these they recommend all-in-one sealed water-cooling systems, usually with a 3 fan rad and of course only for the CPU. GPU and case will be air cooled, although if you mount the rad in the right place it can double as a case fan.
I'm sure there will be air coolers for Threadripper though. Last generation they worked as well as water, the main issue being the shear size and weight and needing to mount it on the CPU.
Re: (Score:2)
Re: (Score:2)
If not, it fails and you wake up to the PC having locked up and the BIOS reporting 80c of thermal overload. Thankfully my CPU survived (3rd gen Core).
a) CPUs haven't died from overheating for years
b) turn off your computer when you sleep. We're screwing the world with our waste enough as it is.
Re: (Score:2)
You can get ones with a 5 year warranty but basically I'd budget for replacing it every 3 years. In fact I'd just swap it out at that point as preventative maintenance.
Re: (Score:2)
Which is a problem. I've used the same cooling kit for my CPU for almost a decade without issue. If you're having to review and replace every 3 years, that's a problem and a significant step backwards.
Re: (Score:2)
It's not great but then again we have 32 cores / 64 threads, base clock 3.7GHz, on an affordable workstation now.
Re: Liquid cooling is a PITA too though. (Score:2)
Why don't you begin at: Why do you need the liquid at all then? If you still need a buttload of fans. You can just put the proper heat sinks right on the components that need it, and flow air through the case with that metric fuckton of 6-12 large slow fans. Same effect, no stupid liquid anywhere. No hassle. No special chipset blocks that only seem to work because you have never actually checked for hot spots in your system.
I never had a leak either. But you still have to unplug water tubes to replace the g
Re: (Score:2)
Running water through a high-power electronics case is not a smart idea. And you will need a fan anyway, and a pump on top, so it's not going to be any more quiet.
Running distilled water is just fine, also modern cooling fluids are inert. You don't need a fan, the pump doesn't have to be on top. The rig I finished building for a friend of mine a month ago doesn't even top out at 15dB under 100% load and was specifically built to be quiet. This isn't 1998 where we were still going to our local machinist and getting blocks of copper cut, and using aquarium pumps to move the coolant.
But the worst bit is that all the stuff on the motherboard won't get any airflow amymore and therefore overheat and die, if you don't add yet another fan right about where a CPU wan would be. So you gain nothing. Except a PITA to install it or swap any components. (No, those no-drip couplings, or whatever you call them in English, aren't 100% drip-free.)
As we all know, cases are 100% hermetically sealed. Never mind that any high heat co
Re: Liquid cooling is a PITA too though. (Score:2)
Running distilled water is just fine,
Just add WaterWetter [redlineoil.com].
Re: (Score:2)
Running distilled water is just fine, also modern cooling fluids are inert.
Err that is a bit of a silly comment to make. If you're into water cooling that you no doubt have seen posts from people frying their systems. Mind you leaks are actually quite rare, but the coolants be they distilled or pre-mixes are far from safe for electronics.
Re: (Score:3)
Re: (Score:3)
Nope, not required.
My 1950x runs just fine with a Noctua NH-U14S-TR4-SP3 heatsink and fan. Note the "-TR4-" in the product code - it's designed for the threadripper.
I was planning to buy a liquid cooler for
Re: (Score:2)
Running water through a high-power electronics case is not a smart idea.
It's something that runs perfectly fine for many thousands of enthusiasts and has for the best part of 20 years. That's before you consider AIO cooling units which are just standard off the shelf parts that are finding their way into every kind of system including some smaller home systems.
And you will need a fan anyway, and a pump on top, so it's not going to be any more quiet.
You should go tell that to all those people who successfully upgraded their high-end air cooling systems to liquid cooling specifically to bring the noise level down.
But the worst bit is that all the stuff on the motherboard won't get any airflow amymore and therefore overheat and die, if you don't add yet another fan right about where a CPU wan would be.
The only parts of the motherboard that need active airf
Re: (Score:2)
Running water through a high-power electronics case is not a smart idea.
Oddly, it's been done for decades in the field of solid state power conversion.
Re: "Optimized for liquid cooling" (Score:3)
Modern high end air coolers can dissipate 300w or more while being quieter than liquid cooling and cheaper. See the entirety of the Noctua lineup.
Re: (Score:2)
Only thing watercooling has over high end air coolers is thermal mass, thus it can handle sudden heat spikes better, but those are quite rare, usually the CPU is mostly idle or doing something for several minutes before going back to idle.
Re: (Score:2)
Actually no. Watercooling has thermal transfer over aircooling. Aircooled systems have thermal mass to boot, and most sudden heat spikes take a long time to actually saturate an air cooler.
As for heatspikes being quite rare, that is also the exact opposite of true. Most modern CPUs clock down massively when idle pushing only a few degrees above ambient. Doing practically anything causes heatspikes, which is why it's so bloody frustrating that we're still doing primitive cooling design such as regulating fan
Re: (Score:2)
Modern high end air coolers can dissipate 300w or more while being quieter than liquid cooling and cheaper.
Generalizations like that will get you into trouble.
There's probably a grain of truth to it, but from a thermodynamic perspective, shipping heat via a coolant to a radiator that is larger than you could conceivable fix on the processor will absolutely yield superior cooling at lower sound volume, and the same amount of airflow can be achieved by a lower RPM fan.
Re: (Score:2)
>that is larger than you could conceivable fix on the processor
This is the part of the equation that you're not understanding. Modern tower coolers have a tremendous amount of surface for forced convection. It's why they can run into three digits USD today, the sheer amount of thin but wide fins interlinked by heat pipes is quite amazing. While you can hypothetically have a greater surface on the exhaust part of a water cooler, most modern water cooler "heat dump" parts are designed to be mounted on one
Re: (Score:2)
Surface area hasn't been a limiting factor for air coolers since the Pentium 4 days. There's a reason why heatpipes are common place now, there's a reason why higher thermal loads require vapour chambers, and there's a reason why both of the above are outperformed by AIOs hitting the market (though I'm waiting for an AIO with built in vapour chamber as the ultimate in current cooling tech.
High tower coolers are great at dissipating heat, but suck at transferring it away from the CPU.
Re: (Score:3)
Modern high end air coolers can dissipate 300w or more while being quieter than liquid cooling and cheaper. See the entirety of the Noctua lineup.
Okay a few things you're missing with your oversimplified comment. The first being the comment about "quiet". No, there's no air cooler on the market that can match the cooling capability of a passive radiator. The ability to cool a computer without spinning fans tops out at very low power CPUs.
Secondly. 300W? What is this, amateur hour? 300W is the amount of cooling provided by small crap 240mm radiators. A normal watercooling system typically provides 500W, with most enthusiasts typically doing dual 360 r
Re: (Score:2)
The G-Line (Score:2)
The other Ryzen 3...G processors are architecturally Ryzen 2... processors with graphics so I wonder what the Athlon 3000G is.
As for me, my previous PC pretty much went belly up after the original Ryzen line was released but before it became widely available. I'd like to buy one but can't justify it to myself at present - in particular because the system I bought back then is not *that* much slower.
Re: (Score:2)
2 non-hyperthreaded Ryzen 2 cores with a Vega 3 unit and 8 PCIe lanes available (8 used onboard by Vega).
I wonder how big my Factorio network can be on it! (Score:2)
I've heard this story before (Score:5, Insightful)
Re: (Score:2)
Unwanted by whom?
One man's bloatware is another's critical feature.
Re: (Score:2)
I Raytrace, Video, and Heavy-Hitter Science Ap (Score:2)
So, obviously, I want one of these. Oddly, I had been trying to hold on to my antique 2600k for as long as I could, hoping that CPU prices would come down.
Currently, it's no fun exporting youtube video that takes hours. I spent over a month re-rendering my Rhino3D resume in Cycles, which mostly ran on the video card, but for those projects too large to run on the GPU, so the machine would run for 36 hours at a time n the CPU.
I did some Paraview visualizations for an optical experiment. The results are one y
What is a "Desktop" CPU? (Score:2)
So what is a "Desktop" CPU anyway? It sounds like a category carefully designed to exclude processors faster than AMD's latest offering. There are low-power processors used for laptops and tablets, and I can see that AMD's latest chip outperforms all of those. But why exclude chips such as the lower end of Intel's Xeon line? You can easily run them in desktop PCs and they are sold for that purpose. Indeed, they don't necessarily cost any more than the latest "enthusiast" parts.
Surely the hardware compa
Re: (Score:2)
The design parameters for high volume high performance server CPUs are different to those for single CPU multifunction desktop CPUs.
If you're doing one thing you can focus on doing it really really well. If you're trying to do everything well you don't have the luxury of sacrificing various things in order to focus.
So yeah, desktop vs other CPUs is a reasonable comparison to make.
Re: (Score:3)
> Single threaded app performance has slowed to a crawl and only a tiny minority of apps benefit from multi-core.
Is this a post from 2005 reposted for our amusement? Because I'll tell you, having even just 6 cores multithreaded (Ryzen 1600) on my desktop is pretty damn awesome. I remember the bad old days when you'd have to shut all your crap off and kill any background processes before doing something as daring as burning a CD, lest you get a buffer underrun and have a coaster. These days with these
Re: (Score:2)
Re: (Score:2)
try to train a tensorflow
That's a good point. That's totally what people use computers for.
Re: Worlds most hyped CPU.... (Score:2)
Re: (Score:2)
about video encoding and compression tools
That's a good point. That's totally what people use computers for.
Look, I'm not saying you're wrong about those applications really needing good parallel processing abilities... But the "modern softwares" thing is kind of silly.
The simple fact is this: 99% of "modern softwares" are going to perform better on 4 5Ghz cores than 847 4.6Ghz cores.
Re: Worlds most hyped CPU.... (Score:2)
Re: Worlds most hyped CPU.... (Score:3)
Re: (Score:2)
I think you greatly overestimate the amount of people who give a shit about such things.
Re: Worlds most hyped CPU.... (Score:3)
Re: (Score:2)
Here are some multi-core CPU benchmarks for encoding video:
https://www.anandtech.com/show/13124/the-amd-threadripper-2990wx-and-2950x-review/10
Am I missing something here? 150 - 250 frames per second is *bad* or something?
Re: (Score:2)
Also, why would I compress a large file with a single thread when 7zip does multithreading no problem? There are benchmarks for that on that link too...
Re: (Score:2)
Re: (Score:2)
Seems it does.
https://imgur.com/a/pUcsAO3
Compressing a 2GB video image. All threads were idle before I started. It's not pinning them all, but that sure looks like a lot more than one thread skating around between cores...
Re: (Score:2)
Tell us some "unspecialized" apps limited by single core performance?
Re: Worlds most hyped CPU.... (Score:2)
Re: Worlds most hyped CPU.... (Score:2)
Re: Worlds most hyped CPU.... (Score:2)
So you're sticking with your original assertion that the majority of users don't use multiple cores, and every example you've been presented with so far is, in your view, an edge case limited to power users.
Ok, try these:
- OS background tasks likes updates or anti malware. Everyone does them whether they want to or not.
- Playing music while doing pretty much anything else.
- Having more than one office app open.
- Pretty much any game made in the last decade. Even console games make use of multi threading.
Hec
Re: (Score:2)
"- Pretty much any game made in the last decade. Even console games make use of multi threading."
They do, but let's remember multi-threading is still a hard task for a programmer to achieve. Besides we've had multi-"cores" for the longest time. We've just called them GPUs.
Re: (Score:2)
This is only true if you are talking about GTA5. We started moving toward multithreaded game engines back at Quake 3. You won't find a purely single-thread-bound game, even made with some Game Maker software
Re: Worlds most hyped CPU.... (Score:2)
So your argument is now "a single core can do what multiple cores can". That's true, but that doesn't mean multiple cores aren't beneficial.
Jeez man, usually when people are flat out wrong and a bunch of people point it out to them, they accept it and move on. You're doubling down on stupid.
Being wrong from time to time is fine. What you're doing is absurd.
Re: (Score:2)
Only thing I've demonstrated is that I know how to set the "foe" setting. You're the weakest link. Goodbye!
Re: (Score:2)
Unfortunately he still has a point when it comes to games. Most of the "oh we totally, really, genuinely support multithreading, HONEST!" games usually run one main thread and a whole bunch of additional threads. And that main thread is the one that will occasionally cap its core out and cause either massive frame drops or microstutter. Which is why single core performance is still king in gaming. No one cares if you have all those extra cores running non-critical threads. It's that main thread capping out
Re: (Score:2)
Single-thread performance has surpassed even the most wild expectations from back when your argument was still theoretically relevant.
Totally depends on use case (Score:2)
The usefulness of more votes vs faster cores is 100% entirely on your use case. For me, even on my phone I had 50 browser tabs open until recently. On my desktop I probably have more. Each tab runs in a separate process, and therefore can use a different core.
I also have multiple VMs, which run on different cores. The most CPU-intensive single thing I'm likely to do is a large compile - which will use as many cores as you have. That, or I'd be encoding to mpeg4 with ffmpeg - which will use several core
Re: (Score:2)
I also utilize all cores with compilation and video transcoding.
However, my Ubuntu VM is mostly idle, except if I compile in one. The many tabs I have open in my browser are all mostly sleeping awaiting input and would have little impact running even on one core. Except for a quick burst on loading where Firefox can now use multiple cores even for one page, and the tab running some multimedia, taking ~5-10% of one core.
I also play GuildWars2 a lot. It is a Dx9 MMO that uses a few cores but is still heavily
Re: (Score:2)
You have very similar use cases to me :)
Multi-threaded compiling + SSD makes a huge difference (of course, much of that is lost by the corporate-mandated protection software which appears to single-thread all disk access ...).
In GW2 I went from an i5 3570 (Ivy Bridge, DDR3 667MHz CL9, boosts to 3.8GHz, realistic 3.6GHz) with GTX 1060 to a Ryzen 5 3600XT (DDR4 3200MHz CL14, boosts to 4.3GHz, realistic 4.2GHz) with the same GTX 1060.
I had been running around 40FPS in most places, with # of character models se
Re: Worlds most hyped CPU.... (Score:2)
Re: Worlds most hyped CPU.... (Score:2)
Re: (Score:2)
Re: (Score:2)
Depends on the application. When I'm building my VM servers for work, 64 cores is ABSOLUTELY better than 32, period.
Re: (Score:2)
Try to run any modern Linux distribution on a single core processor and fell the difference...
Sure, but it's rather intellectually dim to think that because 2 processors is better than 1, 64 must be better than 32.
Nope. Doesn't appear so. Thanks for commenting, though.
Re: (Score:2)
>>Try to run any modern Linux distribution on a single core processor and fell the difference...
>I'm sorry, was your anecdote relevant?
Technically VMWare is a Linux distribution. As is Xenserver. So yes, relevant.
Re: Worlds most hyped CPU.... (Score:2)
Re: Worlds most hyped CPU.... (Score:2)
Some of us do more than play video games. And those of us that do play video games have a browser, YouTube, twitch and a buttload of other things going on at once.
Re: (Score:2)
> Man this is the most BS post I've ever read, single threaded performance is still paramount and you provided NO evidence that "now every app is multi core enabled". You sound like one of those modern gamers who have no fucking clue how computers work.
Except, ya know, I've been a systems administrator since the 90s. Most modern compilers have a switch to attempt to multithread the code they're compiling, and just about anyone with a brain uses it unless they have a specific reason not to. Companies se
Re: (Score:2)
Re: Worlds most hyped CPU.... (Score:2)
Re: Worlds most hyped CPU.... (Score:2)
1998 called, they want you back.
Re:Wishing AMD all the best (Score:4, Insightful)
In fact, I'd quite readily choose to use a SLOWER processor if I could get drastically lower operating temperatures.
But why, when you can choose to use a *faster* processor that will give you drastically lower operating temperatures as well?
Re: (Score:2)
Why do you care if your CPU operates at 70C or 55C?
Aren't you asking the wrong person (i.e., me)? You should reply again to this comment. [slashdot.org]
Re:Wishing AMD all the best (Score:5, Informative)
Your 6 year old Intel Core 2 Quad is jam packed with bugs and security flaws...
Re: (Score:3)
Your 6 year old Intel Core 2 Quad is jam packed with bugs and security flaws...
Every CPU is jam packed with bugs. The erratum for Intel and AMD CPUs often run into the hundreds of pages by the time their generation is over. As for the security flaws. Where? I don't see any. If you mean speculative execution then may I interest you in a course on risk? After all you may not have realised that this largely unpatched flaw covering all OSes affecting the majority of computers in active use in the world is not actually being exploited because a) it's too difficult, b) it's not automatable,
Re: (Score:3)
Yeah, all those browser patches and kernel patches were created for no reason. All just a waste of time. Riiight.
You mean the kernel patches which were off by default and the browser patches that simply fudged the timer accuracy that didn't affect anything right?
That is why those JS in-browser exploits that were demosntrated, and that you already got links to, totally don't actually exist, riight?
Did you try to execute it on your own machine? Of course not. If you did you'd know it doesn't work like that. Not without knowing detailed memory layout of a currently running machine, a layout that is randomised on every execution of a process. If you know that layout what the hell do you need an exploit for?
Just like that IME is /totally/ trustworthy. We swear! --.--
Is that like saying "just like that OS" is /totall
Re: Wishing AMD all the best (Score:2)
Re: The absurdity of claiming to be an atheist (Score:2)
Re: (Score:2)
Re: Rant: Where are the fruits of Moore's Law? (Score:2)
Re: (Score:2)
C is bad, let's use Python!
It's often faster to develop in Python (not a huge fan). I'd rather have the software than not. C is a huge pain in the arse.
Who cares if you copy a string â" it is safer than mucking with pointers!
Three things. Firstly, yes mucking with pointers is a rich source of bugs, which eat up programmer time. Second, C++ obviates the need for most of that without a performance loss. Third, related to 2, the short string optimization means it can be faster to copy since you end up wit
Re: (Score:2)
Exactly! But why bother figuring it out, if the hardware is so fast and large, one can have a separate VM dedicated to each little task?
My point is they didn't manage to figure it out properly 20 years ago when VMs weren't a thing, and haven't now. For some reason many people just can't grasp the basic concepts. 20 years ago you had to have a separate physical machine and that was much much less efficient than a VM.
How many log4j-foo.jar do you have on one computer? They "only" consume diskspace, when not i