Are AMD ryzen3 and pcie 4.0 really the things that I have been holding out since i5 2500 for?
I'm stopping to feel hyped for technology, there's barely any noticeable difference since many years. I doubt even video games did evolve that much, after all, kids still play a game with cubes that look almost like quake 2.
Other urls found in this thread:
cpubenchmark.net
twitter.com
If you have SSD so fats that it's limited by PCI-E then yes, otherwise no.
16 cores in the same TDP at the same frequency ain't bad ngl
When not even Jow Forums can shill something, then there's a reason to not upgrade.
I know this is Jow Forums, but aren't most of applications optimized for 2-4 threads and around the same amount of physical cores, anyways?
You forgot to mention that you need to actually perform operation that is able to utilize such speeds for example rendering some absurdly huge files. There is barely any difference in daily usage between slowest ssd and top shelf nvme drive
No. Current cards don't saturate PCI 3.0. 4.0 is more future proofing then anything, but isn't exactly super useful today or probably for another 2-3 years... for the masses anyways, its a big selling point for datacenters where you may have 8 port NICs on one 16x bus.
The extra processing power in better IPC, clock, and thread count is whats really differences your old Sandy Bridge over Zen2. But with that said, do you run anything that is really CPU intensive? If you mainly just play old game and browse the net then what you've got will likely do you fine for another few years.
Yeah right. It's easy to spot the difference between my 850evo and my 960evo. Heck even the speed with which Firefox opens up. The sata controller adds a lot of lag
If each application is optimized for 2-4 cores, that means you can run one of those applications on each processor with 2-4 cores. If you use a 16 core processor you can now run 4-8 of those applications.
only useful for cloudtards, but intlel uses same fabs for xeon and gaymer
>But with that said, do you run anything that is really CPU intensive?
Yes and no.
Overall, I'm annoyed at my portable devices feeling faster while having celeron and m3 or even snapdragon arm cpus. Also doing some heavy tasks, but only once or twice a week. I imagine there are some things like usb 3 support or even usb c to charge my devices, tablets have build in codecs support, maybe desktops have these too and generally stuff like that.
"Multi-track drifting" with 4 photoshops and a python ide while playing great scott's videos in background? That's a joke.
If thats the case, then on those situations of "Yes" you will notice a BIG difference going from Sandy Bridge to Zen2 or even a Intel 8xx0/9xx0 chip. IPCs have been growing really slowly, but it does add up when you're talking about being 6-7 gens behind. And things like USB 3.1 is definitely nice as well. Considering that RAM prices are slated to drop for months yet (industry got hammered on DDR4), picking up a new system in the next few months or waiting for zen2 (around June or so) is actually a good time to do so luckly.
Heads up in case you do make the plunge, if you ever want to install Win7 on a motherboard that uses USB 3.1 though you do have to prep the ISO first. Asus EZ Installer does the trick and works for any motherboard.
They do saturate pcie3 in multi card configurations. That's why they've used bridges between cards for years
They don't saturate 16x PCI 3.0 even in a multi card (for most cards, it would with 2x Titans granted). The real reason for the bridging interface is because a LOT of motherboards only come with 16 to 24 lanes so when you put to cards in both 16x interface is only working at 8x so the bridge helps with the difference.
Blame Intel on that one for going cheap on PCIE interfaces outside of their Xeon for years.
Yes that's called saturating lanes. Multi gpu set ups involved cloning memory. The Memory bandwidth on the gpu exceeds pcie3 16x bandwidth. Hence the need for a high speed bridge. The amount of lanes and the speed of lanes being limited is down to Intel being stingy, but that is a separate issue
Pcie4 can't come fast enough imo. Imagine an external gpu dock that wasn't hamstrung by thunderbolt 3...
> really the things
Nope.
Ryzen 1st gen wrestles i7 975 on LGA1366 and your CPU as well. cpubenchmark.net
This is literally the revolution of core2 -> core i* magnitude when Sandy Bridge Celerons were as good as Core2 and those who didn't realize that, paid premium for a suddenly obsolete CPU.
Now, being in the same league as Ryzen 3 owners, ask yourself if you need to upgrade. Maybe, but it's good for another 2 to 3 years.
>Blame Intel on that one for going cheap on PCIE interfaces outside of their Xeon for years.
I'm thankful amd gives you enough for a gpu and a nvme drive without having to share any bandwidth
you need to saturate sata, any samsung nvme will do that, but so many other brands fall on their face and are worse then sata options.
a sata that can saturate sata and an nvme are nearly the same cost, so go nvme when you can.
if you only browse the web, its difficult to say the i5 is bad, however consider your tech is nearing 10 years old, and caps on motherboards are typically only guaranteed for 7, mid range amd, the shit you would pay around 200$ for is going to be 8 cores soon, motherboards that support that will cost around 80-200$ though I would get a high end one just because in 2 years you could drop in the highest end amd cpu into it used for cheap.
granted if you're not using all the cores, or doing anything intensive, it's hard to say upgrade, but your motherboard is nearing if not in end of life.
I made a mistake, I meant AMD 3rd gen that might or not come this year, but I didn't mean the AMD's counterpart to i3s. Yet, I can't understand the rest of your post.
>suddenly obsolete CPU
Why will the 3rd gen of ryzens be obsolete soon?
the current 2000 nvidia cards, especially the 2070+ are getting close to saturateing a single pcie 16x and nvme is hitting bottlenecks due to it.
pcie4 allows double the nvme speed, so its helpful for those who can use it, and it future proofs gpus for another few years.
the big difference is only felt when you fully saturate a single core, most applications will never do this.
games put a constant high load on a few cores, so its most felt there, but even still, from sandy to current is a 20% at most ipc gain, which is fuck all honestly.
don't get me wrong, I may going to go from a 1700 to a 3600 or 3500 depending on how things shake out, 8 core to 8 core but without infinity fabric being tied to ram speed. that screams gainz to me, and if it can boost to 5ghz...
>Sandy Bridge Celerons were as good as Conroe Core2
>256k L2/core versus 2mb L2/core
Yeah, nah.
>the current 2000 nvidia cards, especially the 2070+ are getting close to saturateing a single pcie 16x
im just going off what people who do the tests say, I have no real stake in this one way or the other.
>PCIE4.0
Why would a home desktop user care about this?