AMD's budget Ryzen 3300X can outperform an old $350 Intel i7

[GP40X]

They are kidding right?

Looks l;ike Chine is gonna do it to us again.

 

[Starfalcon]

There is a reason they are called corn electronics. Corn goes in the cornhole.

 

[kac77]

You sure it wasn't a b450? "AMD B550 motherboards are expected to be available beginning June 16, 2020".. they still aren't even for sale yet.

It's actually is a B550A. It seems that this model could be actually a 450. Still reading about it.

 

[Ready4Dis]

It's actually is a B550A. It seems that this model could be actually a 450. Still reading about it.

It's an OEM board, still pcie 3.0.. not even sure if it's supposed to be forward compatible or not but I doubt it since it's OEM, meaning it probably wasn't intended to be upgraded and likely wont receive many bios updates, but hey, you never know! It is based on the b450 design and basically just renamed for OEMs (as far as I've seen).

 

[CraigHB]

Nobody wants to put a $120 chip on a $200 motherboard...

Yeah wouldn't make much sense short of a stopgap for a high-end CPU upgrade. I suppose it would be nice to have that upgrade path for the budget CPUs, but B450 motherboards are so much less expensive it's not unreasonable to simply replace the motherboard if required.

That 3300x does seem like a pretty sweet budget CPU. With only one CCX it might actually have better IPC than my 3700x under certain conditions. Just too bad they didn't get the B550 out in a timely manner, they really created a void with that one. Farming it out might have bit them in the ass this time. That's something I think they should have kept in-house.

 

[GP40X]

Nobody wants to put a $120 chip on a $200 motherboard...

I do, but it will be a place holder until Zen 3 and the Ryzen 4000 series chips come out. I am actually torn. Do I go ahead and get a Ryzen 7 3700X and a good board now or get the Ryzen 3 3300X and a good X570 board and see what they have for Zen 3 in the way of a good 4000 processor. later this year. I can't make up my mind, oh the agony, THE AGONY...

 

[Ready4Dis]

I do, but it will be a place holder until Zen 3 and the Ryzen 4000 series chips come out. I am actually torn. Do I go ahead and get a Ryzen 7 3700X and a good board now or get the Ryzen 3 3300X and a good X570 board and see what they have for Zen 3 in the way of a good 4000 processor. later this year. I can't make up my mind, oh the agony, THE AGONY...

Lol, good point, it would make a cheap holdover for people wanting to upgrade later.

 

[travm]

I looked and found some decent deals on X570 boards. Frankly after researching this abit, if i was building a new system today, knowing what I know now, I wouldn't even think of buying a B450, no matter the chip i put in it. If i was $30-40 short on my $800 PC build, I think I could beg borrow or steal that last $40.
The real question is, WHERE ARE THE APU's!
I've been looking to put together a budget system for a friends kid, and its a struggle to get everything inside the budget.

 

[d3athf1sh]

ive never seen intel so down before its lots of fun dont you think LOL

guess you're too young to remember athlon in it's hay day? not the first time intel's been a laughing stock.

hey i've, w/ my personal rig, stuck w/ AMD through it all (cpu's anyway) and really never had any complaints. There's never been a game i couldn't run and just, here recently, finally moved on from my FX-8370 rig late last year. and for what i payed for that proc i wasn't complaining and def got my money's worth. shoot, w/ custom liquid cooling, i was prime stable at 4.82 @ 1.325v. just as w/ all their platforms there's always a way to tweak em a little to get a really nice performance boost. you just have to build it yourself so you have access to bios options. Problem was it usually never was reflected in reviews. (especially launch reviews) just for example phenom's NB OC, FX's NB/HT -> see all the reviews you look at for FX use 2000/2000 for NB/HT which is what was stock for phenom, when in all reality they are suppose to be set to 2200/2600 stock which would smooth out my frame rate big time in GTA V and if you could get the NB higher it would help even more because that controlled not only memory bandwidth but the speed of the L3 cache!!!

But back when i was running FX it wasn't like i was running the top of the line GPU anyway. It did just fine with my 770 and then my (used) 980ti. Which may have left a little performance in the bag, but made it that much sweeter when i jumped to 3800X. and now i can eventually grab an RTX 3080ti in a year or two and still be happy. Man AMD rocks!!!

 

[CraigHB]

guess you're too young to remember athlon in it's hay day? not the first time intel's been a laughing stock

I'm not, that was when I went from Athlon to Core 2. AMD didn't hold the crown as long and it wasn't such a lead. With the process issues it could be a while before Intel regains the crown again. Though Intel being Intel could pull it out sooner than later. In the mean time I'll just be enjoying the good value and good performance with AMD. With the recent reversal on its B450 edict it looks they're still the good guys too. I have an x570 so it doesn't effect me, but still shows a lot of good faith to us lowly retail consumers.

 

[doublejack]

I'm not, that was when I went from Athlon to Core 2. AMD didn't hold the crown as long and it wasn't such a lead. With the process issues it could be a while before Intel regains the crown again. Though Intel being Intel could pull it out sooner than later. In the mean time I'll just be enjoying the good value and good performance with AMD. With the recent reversal on its B450 edict it looks they're still the good guys too. I have an x570 so it doesn't effect me, but still shows a lot of good faith to us lowly retail consumers.

I remember those days. The big dog was the socket 939 Opterons, and AMD was also dominant in the value segment with socket 754 (single channel memory IIRC). I had a socket 754 Sempron that I overclocked in a big way. I remember getting the board and CPU from Fry's for about $90 total. What a value that was.

I think the situation is quite different this time around. Unless Intel has another ace up their sleeve like Core 2 was, which I sincerely doubt because I think they would have played it by now, they are going to take years to catch up to AMD. What Intel did before was take designs they created for mobile platforms, Core 2's intended target, and cross them over to desktop killing Netburst in the process. These days Intel is on a far more unified platform, so there is no alternative design already in the pipeline that they can borrow from. Intel's ring bus does not scale up in cores nearly as well as AMD's chiplet design. Factor in the other areas that Intel has fallen behind on, notably power efficiency, and they are in serious doo-doo. I really don't see any way Intel regains the lead before mid decade, and that's if they start hitting home runs in R&D now. The longer Intel R&D continues to not deliver, the longer AMD will be out front. We'll have a better idea of where Intel is at when Tiger Lake numbers are in and we can compare that to Zen 3/4. I'm anticipating AMD will stretch out the lead over the next couple CPU gens.

 

[travm]

So Toms hardware is reporting AMD may allow some B450 and X470 motherboards to upgrade to Zen 3. Looks like a poor solution, but there may be an upgrade path.
maybe they'll add my X370 to the list!

I remember those days. The big dog was the socket 939 Opterons, and AMD was also dominant in the value segment with socket 754 (single channel memory IIRC). I had a socket 754 Sempron that I overclocked in a big way. I remember getting the board and CPU from Fry's for about $90 total. What a value that was.

I think the situation is quite different this time around. Unless Intel has another ace up their sleeve like Core 2 was, which I sincerely doubt because I think they would have played it by now, they are going to take years to catch up to AMD. What Intel did before was take designs they created for mobile platforms, Core 2's intended target, and cross them over to desktop killing Netburst in the process. These days Intel is on a far more unified platform, so there is no alternative design already in the pipeline that they can borrow from. Intel's ring bus does not scale up in cores nearly as well as AMD's chiplet design. Factor in the other areas that Intel has fallen behind on, notably power efficiency, and they are in serious doo-doo. I really don't see any way Intel regains the lead before mid decade, and that's if they start hitting home runs in R&D now. The longer Intel R&D continues to not deliver, the longer AMD will be out front. We'll have a better idea of where Intel is at when Tiger Lake numbers are in and we can compare that to Zen 3/4. I'm anticipating AMD will stretch out the lead over the next couple CPU gens.

Recall Intel picked up Jim Keller. The architect of both of AMD's massive successes.
iirc, time from hire to $$$$$ on Ryzen was in the neighbourhood of 3 years. Someone else who keeps better tabs on this stuff or like doing pointless research can likely narrow that down or actually pick the correct date.

Intel will be back swinging. But CPU design cycles are multiple years. I'm more concerned with what AMD has coming after ryzen, because i'm not sure they have the depth in R&D to pull off a repeat. That said the Ryzen design likely has another 5-10 years left.
None of this can be backed with technical information, just my perspectives based on memory of what happened in the past 20 years.

Also from my interpretation of all the news the past 5 years, intel kind of bogged themselves down in FAB struggles, lost focus maybe. I think they are back, fully laser focused. But it will be 2-3 years before we see the light.

 

[d3athf1sh]

AMD didn't hold the crown as long and it wasn't such a lead.
With the recent reversal on its B450 edict it looks they're still the good guys too. I have an x570 so it doesn't effect me, but still shows a lot of good faith to us lowly retail consumers.

actually amd had taken the lead with athlon xp but when athlon64 came out they were blowing P4 out the water. not to mention they were easier to cool. And all that was with a lower clock speed. P3&4 was intel's bulldozer. core 2 was ok but once phenom came out it they were back ontop. Core Ix was the beginning of intels most recent reign but when you look at them now with all the security holes patched they aren't anywhere near as fast as they use to be. Just think how much different things would have been if Intel wasn't cheating by building insecure processors.

Intel will be back swinging. But CPU design cycles are multiple years. I'm more concerned with what AMD has coming after ryzen, because i'm not sure they have the depth in R&D to pull off a repeat. That said the Ryzen design likely has another 5-10 years left.

yeah and they had to go back and start a new design from scratch because of security issues. core series is played out at this point anyway. thought it was funny when i started seeing I7 laptops with a dual core proc's in them, lol. If Intel keeps up their anti-consumer schemes there gonna lose even more of their loyals. What's scary is prob 90% of the worlds servers and data centers are slam full of intel gear.

Well hopefully amd can get there money's worth from ryzen, an 8 core should be good enough for games for a few years anyway. Just hope they don't use ryzen as a crutch tho and keep up their R&D. Like you said Intel's down but not out. But anyone that buys core architecture at this point is askng for it plus they need to learn a lesson.

 

[Ready4Dis]

So Toms hardware is reporting AMD may allow some B450 and X470 motherboards to upgrade to Zen 3. Looks like a poor solution, but there may be an upgrade path.
maybe they'll add my X370 to the list!


Recall Intel picked up Jim Keller. The architect of both of AMD's massive successes.
iirc, time from hire to $$$$$ on Ryzen was in the neighbourhood of 3 years. Someone else who keeps better tabs on this stuff or like doing pointless research can likely narrow that down or actually pick the correct date.

Intel will be back swinging. But CPU design cycles are multiple years. I'm more concerned with what AMD has coming after ryzen, because i'm not sure they have the depth in R&D to pull off a repeat. That said the Ryzen design likely has another 5-10 years left.
None of this can be backed with technical information, just my perspectives based on memory of what happened in the past 20 years.

Also from my interpretation of all the news the past 5 years, intel kind of bogged themselves down in FAB struggles, lost focus maybe. I think they are back, fully laser focused. But it will be 2-3 years before we see the light.

Yeah, their next core is already designed so likely 3+ years before anything he is involved in sees the light of day. Hopefully he isn't stuck working on a new architecture for 14nm .

 

[doublejack]

Yeah, their next core is already designed so likely 3+ years before anything he is involved in sees the light of day. Hopefully he isn't stuck working on a new architecture for 14nm .

And that's what I'm saying. 3+ years for anything to see the light of day puts us into the 2024-25 time frame, and I said mid-decade AMD is going to be on top for a while. It could conceivable take even longer for Intel to make up the gap because it is a moving target they're trying to hit. AMD will have years of additional R&D to stay ahead.

 

[Aspendancer]

I'm more excited about still being able to drop these chips into my old X370.

Me too.

 

[travm]

And that's what I'm saying. 3+ years for anything to see the light of day puts us into the 2024-25 time frame, and I said mid-decade AMD is going to be on top for a while. It could conceivable take even longer for Intel to make up the gap because it is a moving target they're trying to hit. AMD will have years of additional R&D to stay ahead.

Remember, the last time (before Ryzen) AMD's years of R&D produced Bulldozer.
there is quite a bit more to R&D than just time. And i'm afraid Intel has more of all those other things.

 

[Ready4Dis]

Remember, the last time (before Ryzen) AMD's years of R&D produced Bulldozer.
there is quite a bit more to R&D than just time. And i'm afraid Intel has more of all those other things.

More != Better in all cases. Get to many engineers together with to many good ideas and an unlimited budget and you end up shooting for 10nm that isn't reasonable instead of a tamer 10nm that would have actually worked and kept you on top. There are lots of instances of smaller less funded entities outperforming others be it in business or sports. The resources do give you staying power and some advantages, but it doesn't automatically mean you'll win. And on the same token, one great CPU architecture from AMD doesn't mean another bulldozer won't happen either. I know I had you're saying though, resources do help I general with having more ideas and being able to test more things.

PS. I believe AMD probably had more R&D resources for bulldoze than they did for ryzen given the success of their previous chips vs the budget they were able to maintain selling bulldozer before zen.

 

[ManofGod]

Remember, the last time (before Ryzen) AMD's years of R&D produced Bulldozer.
there is quite a bit more to R&D than just time. And i'm afraid Intel has more of all those other things.

Not really sure how to interpret what you are saying here. AMD is not the pre Bulldozer AMD of yore but the all new AMD with a knowing what they are doing and doing it person at the helm.

 

[travm]

Not really sure how to interpret what you are saying here. AMD is not the pre Bulldozer AMD of yore but the all new AMD with a knowing what they are doing and doing it person at the helm.

I agree. Exciting, i cant wait to see what comes next. in 2-3 years.

 

[OFaceSIG]

It's kind of crazy you still can't buy these. Looks like I might be booking a 1600AF for my kid.

 

[thesmokingman]

PS. I believe AMD probably had more R&D resources for bulldoze than they did for ryzen given the success of their previous chips vs the budget they were able to maintain selling bulldozer before zen.

I'm not sure... that is they did have more resources but they did not develop BD in the traditional way. Under Ruiz who ran the company to the ground, they had the brilliant idea to reduce/remove engineers during the Bulldozer optimization period. They did this by applying GPU design principles to Bulldozer. Basically with GPUs you can use software during fine tuning of circuits etc whereas with CPUs they are more done by hand. Software optimization speeds up this process obviously at the cost of efficiency. Back then there was a mass exodus of engineers out of AMD. Ruiz was letting egineers go left and right. The Zen architecture on the other hand was designed with a skeleton crew led by Keller. Keller however isn't the principle designer you know right? Keller is like the super team manager. I think they had like 400 ppl on the team. Compare that to Keller's silicon engineering team at Intel which has over 1k ppl, and that's just process team lol.

Also, you are forgetting that they really had no profits because they were being frozen out of the OEM markets by Intel. Hector Ruiz was burning the company down at the time, all the divisions sold, the overpaying for ATI, on and on. They were not in a happy place. Then again when they unleashed the their engineers to do whatever they wanted with Zen, AMD was near bankruptcy. I'm not sure which was a worse condition in general, but Su had the company on lockdown when she got the Helm. AMD had already fallen there was no where to go but up or death. All their bets payed off from Mantle, to the consoles, and underpinning all that Zen.

https://www.statesman.com/NEWS/2017...tin-engineers-are-helping-AMD-challenge-Intel

 

[Ready4Dis]

I'm not sure... that is they did have more resources but they did not develop BD in the traditional way. Under Ruiz who ran the company to the ground, they had the brilliant idea to reduce/remove engineers during the Bulldozer optimization period. They did this by applying GPU design principles to Bulldozer. Basically with GPUs you can use software during fine tuning of circuits etc whereas with CPUs they are more done by hand. Software optimization speeds up this process obviously at the cost of efficiency. Back then there was a mass exodus of engineers out of AMD. Ruiz was letting egineers go left and right. The Zen architecture on the other hand was designed with a skeleton crew led by Keller. Keller however isn't the principle designer you know right? Keller is like the super team manager. I think they had like 400 ppl on the team. Compare that to Keller's silicon engineering team at Intel which has over 1k ppl, and that's just process team lol.

Also, you are forgetting that they really had no profits because they were being frozen out of the OEM markets by Intel. Hector Ruiz was burning the company down at the time, all the divisions sold, the overpaying for ATI, on and on. They were not in a happy place. Then again when they unleashed the their engineers to do whatever they wanted with Zen, AMD was near bankruptcy. I'm not sure which was a worse condition in general, but Su had the company on lockdown when she got the Helm. AMD had already fallen there was no where to go but up or death. All their bets payed off from Mantle, to the consoles, and underpinning all that Zen.

https://www.statesman.com/NEWS/2017...tin-engineers-are-helping-AMD-challenge-Intel

I didn't suggest they used their resources wisely, just that they probably had more available at the time. Resources != Results if they aren't handled properly. As you can see having as many engineers as Intel does did not ensure 10nm would b successful... It merely allows them more mistakes before it would otherwise have bankrupt other companies.

 

[lopoetve]

actually amd had taken the lead with athlon xp but when athlon64 came out they were blowing P4 out the water. not to mention they were easier to cool. And all that was with a lower clock speed. P3&4 was intel's bulldozer. core 2 was ok but once phenom came out it they were back ontop. Core Ix was the beginning of intels most recent reign but when you look at them now with all the security holes patched they aren't anywhere near as fast as they use to be. Just think how much different things would have been if Intel wasn't cheating by building insecure processors.

Eh, as someone with a background in compute, that's not quite accurate - some of those holes applied to AMD too (and ARM, and others). Speculative execution was lauded at the time as enabling better performance, no one thought you'd be able to actually weaponize exploits for things like Spectre/rowhammer/etc. That blew apart a whole conceptual path of compute design - while it's their fault that it happened, it's not like we were sitting there for a decade warning that it might happen either - no one thought it really would.

yeah and they had to go back and start a new design from scratch because of security issues. core series is played out at this point anyway. thought it was funny when i started seeing I7 laptops with a dual core proc's in them, lol. If Intel keeps up their anti-consumer schemes there gonna lose even more of their loyals. What's scary is prob 90% of the worlds servers and data centers are slam full of intel gear.

Well hopefully amd can get there money's worth from ryzen, an 8 core should be good enough for games for a few years anyway. Just hope they don't use ryzen as a crutch tho and keep up their R&D. Like you said Intel's down but not out. But anyone that buys core architecture at this point is askng for it plus they need to learn a lesson.

It's a CPU; many consumers just don't care. At the end of the day, if it does the job required, mission accomplished - for those in the know, Ryzen certainly wins right now, but almost no one at scale cares about which CPU they're using unless they need ultra-dense compute (Epyc 64 core, for instance). I love Ryzen, but if someone handed me a 10900k, I'd use it - it'd do the job. Or if it cost $200. For my own money? I'm going AMD again. Intel will be back though. This is cyclical, we'll be here again.

 

[lopoetve]

I'm not defending Intel's actions. They have a tendency to replace their motherboards and chipsets far more than is actually necessary. However, there are positive things that result from that. The people that complain about changing motherboards are often the same people who come into the forum and make threads about replacing their CPU and not being able to flash it, or asking about AMD CPU compatibility, because its not as cut and dry as they thought it would be. These same people also find that there may be additional limitations of memory clocking or that their old budget boards don't support some of the features that the new CPU's do.

When we are talking about $150 to $200 motherboards, I generally couldn't care less about having to replace the board to get the best possible computing experience and the most out of my new processor. When you are talking about $600+ motherboards, then I think we are well into constant motherboard changes being an issue. Of course, the more expensive motherboards didn't make as many compromises in their designs and using a newer CPU with a high end motherboard that's a generation old often comes with fewer problems. The weird thing is that the people who buy the ultra-cheap motherboards tend to be the ones that don't want to replace their boards. The ones who have the $600+ motherboards are often the ones who will replace it each time there is a new chipset to get the latest features, the best memory clocks, and the best overclocking.

But when you change motherboards and chipsets each generation, you avoid the pitfalls of requiring an old CPU to even get the system to POST with a newer CPU. You avoid your BIOS ROM being to small to support a broad range of CPU's. You get up to date VRM designs and implementations that generally allow for better overclocking of modern CPU's. You also get the ability to clock RAM higher than you might on the older motherboards with those newer CPU's. Modern power savings features often don't work on legacy motherboards. The list of disadvantages goes on and on. Intel broadly avoids all that crap. If you match up the socket with the CPU you are normally good to go. Intel's had broad socket compatibility in the past and the same problems AMD's platform has were all very much present. It has also had artificially short life spans for chipsets and sockets and occasionally people have found enough evidence to call them out on that.

You pick your poison, but whether you were tired of buying new motherboards or not, there are plenty of valid reasons to go either direction. Personally, even with AMD's broad socket compatibility I opt for newer motherboards. I've even replaced the motherboard without replacing the CPU in order to gain access to newer platform features when there was something there I wanted. Of course, this was largely back when Intel's CPU's were stagnant and there was far more innovation on the motherboard side. These days CPU's are moving forward again on the AMD side, so it's largely had reasons to upgrade each generation while Intel hasn't.

This. I stopped caring about minor improvements (which most cpu upgrades have been; AMD's big jumps notwithstanding - you ain't gonna notice going from a 6700k to a 7700k, for instance), so when I do upgrade, I want it all. I also hated the flash cpu dance and borrowing / buying crap chips to do BIOS flashes when a compatible bios, well, wasn't.

But I also only upgrade every 3-4 years. E8400 -> 2600k -> 6700k for gaming, Phenom II -> FX-8350 -> 1700 for workstation. I'd be on the 1700 for a lot longer if I didn't have an urge to jump to the latest right now.

 

[Bankie]

The key to the speed boost of the 3300x is that there's a single CCX in the CPU, which has a significant impact on reducing latency between cores. GN found that it makes up to a 20% difference clock for clock versus the 3100.

Whether the 3300x is faster than a 3600x comes down not to clock rates but whether a game can use more cores. If you watch the GN video you'll see their examples of this in action. Also, it is quite an exaggeration to say a 7700k is way faster. Even at 5.1GHz there's at best a small advantage over a 3300x at stock speeds. In many cases they are effectively equivalent.

This is actually why this chip is a 7700k killer. It has the same core and thread count, and better IPC. In games where more cores matter then neither of these CPUs is in the top tier. Note that people with a 7700k shouldn't sell their chip and run out and buy a 3300x. It is just that anyone looking for performance in that tier now has a no-brainer option. Just buy the 3300x and call it a day. Used 7700k chips should now have a value of less than $100.

It's not a "7700k killer" though. Yes it's a "little bit better" at productivity but when gaming the 7700k is generally anywhere from 3600x class to a "little bit better" than a 3900x.

The 7700k @ stock was often about equal to an 8600k for gaming purposes. The TPU benchmarks show that overall the 8600k is 8% faster at stock at 1080p and 2% faster at 1440p than the 3300x.

https://www.techpowerup.com/review/amd-ryzen-3-3300x/15.html


GN recently did a 1700x follow up and include the 7700k in many of the gaming benchmarks which shows it outperforming all of the AMD CPUs quite often.




When we're talking about performance differences of less than 10% between the best and the worst it does seem kind of silly but I do wonder what's going to happen once we have GPUs faster than the 2080ti; it's possible that the old and busted 7700k's lead will increase even more.