add arrow-down arrow-left arrow-right arrow-up authorcheckmark clipboard combo comment delete discord dots drag-handle dropdown-arrow errorfacebook history inbox instagram issuelink lock markup-bbcode markup-html markup-pcpp markup-cyclingbuilder markup-plain-text markup-reddit menu pin radio-button save search settings share star-empty star-full star-half switch successtag twitch twitter user warningwattage weight youtube

Opinions on AMD'S Computex Keynote

adeoluy2k

1 month ago

Opinions on AMD'S Computex Keynote

Comments

  • 1 month ago
  • 2 points

Nothing of real substance was shown CPU side leaving everyone as much in the dark as before for all the hype.

Navi is disappointing pulling just ahead of a Vega 64.

Gotta wait for E3 to get more information if they don't keep everything under wraps until NDA lifts.

  • 1 month ago
  • 1 point

that's the plan I'm waiting for the e3 reveal and the price drops which will make my choice easy

  • 1 month ago
  • 1 point

I wouldn't bank on pricing drops AMD is pricing the new models well over what first and second generation parts are going for now so things are pretty fair as it stands.

2600 is $150-$160 well the 3600 is $200.

2700X is $280-$300 well the 3700X is $330.

  • 1 month ago
  • -3 points

when rtx dropped gpu prices fell dramatically initially if that happens again it wil benefit those who don't think navi is worth it

  • 1 month ago
  • 2 points

GPU pricing didn't drop on NVidia side during the RTX launch the cards they replaced were mostly all sold out before launch.

AMD did a massive pricing cut to try and stay competitive. But with the NVidia counterparts already likely ahead still that situation isn't likely to be reversed.

If We go by AMD's Demo the Navi is only 3-5% ahead of the Vega 64 and with the Sapphire Rep letting it slip that its MSRP is $399/$499 depending on the model that just isn't worth it whether it is the flagship or cut down model.

  • 1 month ago
  • 0 points

in Canada they did i was on amazon the day after and everything dropped it ddnt last long but there was a drop

  • 1 month ago
  • 1 point

I think the 3000 series is gonna knock down team blue for a while. LITERALLY ALL of them can boost up over 4 GHz. AMD is stacked right now. But I want to see how Intel will respond.

  • 1 month ago
  • 1 point

If intel responds regardless of the outcome price is going to matter alot and ryzen hasn't revealed the lowe end stuff yet

  • 1 month ago
  • 1 point

Either way AMD is in the lead

  • 1 month ago
  • 1 point

Ryzen 3000 will outsell Intel among people that build PC's for personal use. This is a small part of the Intel business pie. A scratch, not a Mike Tyson KO.

  • 1 month ago
  • 1 point

There is already increasing interest among OEMs, including in the server space. Not a KO, but I'd expect AMD to continue increasing market share across all markets.

  • 1 month ago
  • 1 point

I was pretty hyped until I thought about what was actually said. First they are releasing 2 Ryzen 7's with one being $399 that they are saying is going to compete with Intel’s i9-9900k that was released a year ago for $488. Is this really a win and it doesn’t include an iGPU? Say what you will about iGPU but it's still a perk. The 12 core chip is definitely cool. I guess we will just have to wait and see.

  • 1 month ago
  • 1 point

I think it's intriguing, but until we see some independent benchmarks / reviews of the various Ryzen skus it's still all hype.

That being said, I'm more confident that Ryzen 3000 will end up being more relevant in relation to Intel's offerings, than Radeon 5700 will be to those from nVidia.

  • 1 month ago
  • 1 point

I'm more excited for gpu price drop for older stuff like the vega 56, 64, and rx series.

  • 1 month ago
  • 1 point

this is what im also looking foward to

  • 1 month ago
  • 1 point

AMD's flagship Ryzen 9 starts at $499 - this will make Intel fall to their knees. If/when Intel releases a new 12-core, they'll have to price it at like 700-800 with higher clockspeed to compete with AMD's pricing. But even if that's the case, I don't think it would be worth the extra $200-300.....

(But the question is: Will Intel even price a 12-core at that price? lol)

Of course, even though it is 2019, 12-cores is overkill and unnecessary for the majority of PC users, so right now the major CPU battle will be between the 8-core SKUs from AMD and Intel.

To think that the standard core count on the mainstream units has risen to 8 cores in 2 years is mind-blowing to me. I'm still running a 6700k, and at the time that I bought it, I thought we would be stuck with 4C/8T for at least a decade longer because of Intel's dominance in the market.

But then AMD ultimately released Ryzen and I was blown away. With Ryzen's release, we basically skipped a 6-core standard (even though now mainstream units have 6C/12T) and went straight to an 8-core standard. That was just incredible.

And now, I think we may be entering a 12-core standard with this new Ryzen-3000 launch, which is absolutely astonishing. Granted, of course, $499 ain't cheap, and based off of previews of the x570 MOBOs, those won't be cheap either. But with time, I think the new MOBO costs will decrease as the manufacturers find better and more efficient ways of integrating PCIE 4.0 onto the platform, so I fully expect MOBO prices to go down as time goes on.

It's actually getting scary for Intel. Their financing and business management departments are gonna have to rethink their pricing strategy to keep up with AMD.

Because if AMD releases a flagship 16-core (not Threadripper) for like $600-700 dollars that's based off Zen 2, then IDK what Intel is going to do.

That aside, Intel's 9900k will still be good, and even with AMD's demo at the event, I still think the 9900k will be the best 8-core gaming CPU on the market. But if its price stays at $480-500+, I honestly think the Ryzen7 3800X would be a better buy, TBH. Newer technology and at the very least it's part of the PCIE 4.0 platform.

(Let's be honest though, just spend the extra $100 to get the 12-core 3900x.)

The only real problem that may come to light is heating issues with the MOBO due to the new technology. But again, I'm confident manufacturers will overcome this even if problems do arise. Hopefully, there won't be any issues at launch though.

AMD has gained so much momentum in these last 2 years. If Intel keeps remaining stagnant, they might just become the old AMD.....

Also, INB4 AMD releases 20-core/24-core flagship by 2020 (or 2021). 20 cores or 24 cores become mainstream by 2020/2021. Lel.

Now, THAT would be scary. And oh god, at this rate, it's actually plausible.... (Even though it's unnecessary.)

  • 1 month ago
  • 1 point

Underwhelming.

I really had high hopes they'd make 6 cores, 12 threads entry-level. I mean, 4 cores, 8 threads is close to the end of its days for gaming at this point. Yeah, sure, they're still fine and I wouldn't upgrade if I had one, but for newer games, their days are limited. I can't imagine a 4 core, 4 thread CPU for Ryzen 3, that would just be absurd... But who knows? I suppose I was expecting Ryzen 3 to be 6 cores, 12 threads, Ryzen 5 to be 8 cores, 16 threads, Ryzen 7 to be 12 cores, 24 threads and Ryzen 9 to be 16 cores, 32 threads, if it exists. That would be far more appealing and will absolutely shoot Intel out of the market. If this was the case, it'd be safe to say by the end of the year, AMD should hold more than 30% of the market share.

I wasn't expecting 5.0 GHz from the CPUs, because that's beyond unrealistic, but was a 10% clock speed improvement too much to expect? 3700X only has a boost clock of 4.4 GHz, which is clearly single core, so all-core is probably around 4.2 GHz, which is only 200 MHz above 2700X's 4.0 GHz. The 15% IPC improvement is probably where substantial performance kicks in, but even then, it's hard to judge. There's also no word on latency, which was the biggest prevention of Ryzen CPUs from overclocking well so far. If latency hasn't been improved, that means what we see will pretty much be what we get, there's no hope for more performance from overclocking.

3800X is shrouded in mystery. Boost clock speed is 4.5 GHz as opposed to 3700X's 4.4 GHz. Costs $70 more, uses 40W more power. This makes absolutely no sense on paper, price to performance/value for money is way lower on this than the 3700X. Why would anyone get the 3800X if it's only 100 MHz faster and uses so much more power, meaning it'll be a lot warmer too? No idea, but hopefully AMD has something up their sleeve with this.

Honestly, I don't expect any of these CPUs to be better than the 9900K in gaming for most games/on average. The charts were trying to say they have beaten the 9900K in both single core and multi core performance, but I bet those charts were using the 9900K at stock speeds. Overclock it to 5.0 GHz and it's a different story. Given that the 9900K is faster than the 2700X by 20% - 30% depending on the situation, I'm probably going to guess the 9900K will be about 10% faster than the 3700X still. Very minute, so it's probably not worth the jump in cost to it any more, even if someone did want "the best."

Navi... In all honesty, not sure how I feel about it. Always take demonstrations with a grind of salt. They used an AMD optimised game to showcase it's performance, so obviously it was going to do better on the Navi card than the 2070. I'm surprised they didn't use World War Z... We also don't know what resolution they did it at. AMD used 4K to show the 2700X having matching performance with Intel before... Yeah, 4K, GPU bottlenecked, didn't show anything about the 2700X really. I'm holding the same grind of salt against this. If these aren't priced competitively against Nvidia, i.e. cheaper, there's going to be no market for them. Radeon VII set a good example; same price as the 2080, slightly behind in performance, loud, warm, uses more power, so even with free games, it just didn't sell. Will Navi suffer the same fate? Hard to judge. Given that RX 580 couldn't be judged properly at the time of launch because it launched during the stupid cryptocurrency mining boom, it's hard to predict how the Navi will do. However, post the mining boomb, the RX 580 was a far better deal than the 1060. If Navi launches to be a better deal than the 2070, 2060, 1660 Ti and 1660 at all the price points, it'll sweep the mid/low end market for sure. If not, Nvidia will keep on reigning.

All-in-all, I was slightly disappointed, but only slightly. AMD should still be the value king, so there'll be even less reason to recommend Intel now. Hopefully by E3 or at the latest, July 7th, second generation CPUs will see a dip in price. If the 2700X goes for $250, it'll be a better value than the 3700X judging by the current numbers, but we'll need to wait for benchmark numbers to know for sure.

  • 1 month ago
  • 4 points

I suppose I was expecting Ryzen 3 to be 6 cores, 12 threads, Ryzen 5 to be 8 cores, 16 threads, Ryzen 7 to be 12 cores, 24 threads and Ryzen 9 to be 16 cores, 32 threads, if it exists. That would be far more appealing and will absolutely shoot Intel out of the market. If this was the case, it'd be safe to say by the end of the year, AMD should hold more than 30% of the market share.

AMD is playing the game. They could have released a lineup exactly as you have described if they wanted to. They have the tech ready to do that.

AMD knows that there is no upside to leaping ahead of the market like that. Doing so will result in a burst of sales volume that they would not be able to fill, followed by a lull of stagnation both in sales and improvement for many years to follow.

Playing it steady, pricing new products into the existing market, participating in the train of incrementalism is the long game and it wins the race for both AMD and Intel.

You think AMD or Intel wants either one "shot out of the market?" It doesn't do either of them any good to wipe each-other out. The moment one of them files bankruptcy an antitrust judgement will split the other in half. There's no benefit to trying to sorely beat the competition here. Intel stopped making leaps when AMD fell behind, because they knew they had no choice if they wanted to keep their company in one piece.

  • 1 month ago
  • 1 point

You see, I totally understand that, and that's why if it were me, I wouldn't prepare a 16 core, 32 thread CPU (apparently confirmed and announced at E3, but I'm still dubious about this) for third generation and save it for fourth generation, for that improvement and increment you speak of.

The whole point of making Ryzen 3 6 cores, 12 threads is purely to steal market share. Despite two years of Ryzen being superior in value, they still only hold 20%+ market share, so I would've wanted to take a jab on Intel while I could to even the market as much as possible before I hit AM5. I say shoot Intel out of the market, but that only applies to the current CPUs (or maybe next generation as well) so it gives AMD more time to work on things they still currently lack behind in and get the revenue they need. Market share is just too crucial for long-term revenue, and if they don't grab it while they can, when Intel catches back up, it's just going to be difficult to do it then. Despite two years of better value, Intel is still the preference of many by default, so if Intel catches back up in what they offer, I can only imagine the market being fixed at best, if not slowly declining for AMD, due to more support available from Intel as they have more users. Of course, this is assuming Intel can make it back in the next two years. If AMD feels like Intel won't be able to and they can even things out in 2+ years, then yes, of course this is the better route, but I would've preferred the safe route, just in case Intel does manage it.

For me, a 16 core, 32 thread consumer level CPU is absolutely unnecessary right now, not even to further game development. When Ryzen was first introduced and gained popularity, it quickly pushed games to develop in need of more cores and threads, this was great, but the games that require this are still few. And because game companies want to sell as many copies as they can, they're not going to leap too far ahead, i.e. they'll work towards 12 cores, 24 threads, but not as fast as they worked towards 8 cores, 16 threads, simply because when 8 cores, 16 threads came out, they were stuck on 4 cores, 8 threads for like a decade, so the market was already "full" and ready to ask for more. Now that more exists, not the entire market is on the same level, it's not nearly as "full" as it was two years ago, with a lot of people sat on 6 cores, 12 threads. Besides, I have very high suspicion that RAM might be the bottleneck for 16 cores, 32 threads CPUs. Still being dual channel memory, this is an issue I see, as previous 16 core, 32 thread CPUs, the Threadrippers, were using quad channel memory. I'm not an engineer or a professional, so we'll have to hear from people with more knowledge on this matter about that.

  • 1 month ago
  • 1 point

I would like to see the new Ryzen 3 series flush out with 4C/8T or 6C/6T only. Saving 4C/4T for the next Athlon entry product... I think that would make a pretty well rounded product stack.

on the subject of the memory bandwidth issue... It is probably an issue, there's definitely evidence to suggest a performance leap going to quad channel platforms in some workloads even when using similar core count CPU's.

I bet the doubling of L3 cache will help alleviate some of that issue. Increasing memory clock speed support helps as well.

  • 1 month ago
  • 1 point

At this point, yeah, I can't imagine another 4 core, 4 thread CPU to be Ryzen 3. I mean, announcing that at E3 is going to be more underwhelming than this, so I'd save that for the next Athlon too. 200GE faired pretty well, considering it was overclockable, so if they make the next Athlon 4 cores, 4 threads, overclockable, it's going to have a decent market.

Yeah, but these aren't really fixes to the issue. I'm surprised they let this slide when one of the biggest criticisms for their 2990WX, 32 core, 64 thread CPU, was the issue of memory being quad channel, substantially limiting what the CPU was actually capable of. I mean, L3 cache and higher RAM speed support do alleviate it, but it feels like using duct tape for a leaking tap; it's not really the solution to limited memory bandwidth. Truth be told, 3200 MHz RAM is currently the deal, and getting higher clock speeds increases latency (unless you get the ridiculously expensive b-die RAM) so it's kind of meh. I suppose it doesn't quite make sense to make the entire X570 (and B550 if they exist) line quad channel to further increase the motherboard cost, given it's gone up "substantially" having to be better quality and have PCIe 4.0, but would be weird if some were quad and some were dual, and that's why I would've held off the 16 core, 32 thread CPU for third generation. Bleh, I don't know. I suppose they know they've won this year, so it didn't matter, and now after Intel's speech, this is reassured.

  • 1 month ago
  • 2 points

Keep in mind a lot of the performance problems on the 2990WX were actually caused by the Windows scheduler totally botching the way it was scheduling and moving work around on that CPU. Many of the same workloads that saw abnormally low performance on Windows with the 2990WX, scaled up perfectly fine in Linux.

  • 1 month ago
  • 2 points

Sounds more like you just have trouble managing expectations.

  • 2017 - 12 core 1920X launched for $800 and required an X399 board.
  • 2018 - 12 core 2920X launched for $650 and required an X399 board.
  • 2019 - 12 core 3900X is launching for $500 and doesn't require an expensive board.

To be frank, I don't think AMD cares if you're "underwhelmed".

  • 1 month ago
  • 0 points

I do, frankly, I was expecting a little more than this.

And I disagree that it "doesn't require "an expensive motherboard." With the PCB improvement, PCIe 4.0, cooling necessary due to the higher power, etc, this won't be far off from the X399 (at least the higher end ones that allow for better power delivery), if your basis for "expensive" is that the X399 is expensive, and I'm not including the ridiculous limited edition motherboards, such as ASRock's liquid cooling limited 999 production motherboard.

They don't, I never said they cared. The post was asking for opinions, and I'm stating mine.

  • 1 month ago
  • 1 point

this won't be far off from the X399 (at least the higher end ones that allow for better power delivery),

You mean like Z87, Z97, Z170, Z270, Z370, X370, X470, and so on?

They've all had their share of $300+ motherboards.

  • 1 month ago
  • 1 point

No, just something equivalent to their past counterparts in the high end that has good VRMs to support the 12 and 16 core CPUs, but not ridiculous, e.g. Your usual Gaming 7 (or now Master), Taichi, etc. Nothing ridiculous like the Xtreme, Extreme or Godlike that just weren't designed for the average consumer.

[comment deleted by staff]
  • 1 month ago
  • 1 point

But then you might as well shift all the way back to B350 if you want it to "just work." And honestly, an X470 motherboard with good VRMs to support the 3900X well is not that much cheaper than an X399 right now.

Looking at the PRO CARBON, Taichi, Gaming 7, etc, you're looking at $190 - $210 already. An X399 is about $300 (excluding budget ones and discounts). So assuming the X570 to be cheaper than the X399 by a little, but more expensive than the X470, a guesstimate would be $250 - $270 minimum for a decent X570. Are you really going to save $50 and get an X470 over an X570 when you're buying a $500 CPU that will clearly be paired with high end parts? $50 is going to work out like 3% of your overall budget. Granted, if I had a nice X470 already, I would not swap to an X570, but then again, is there a need for the upgrade from a second generation to third generation? If so, how many people need this upgrade?

Furthermore, buying an X470 for it will have the same issue second generation had with B350/X370 motherboards; BIOS update with a previous generation CPU before the motherboard will recognise the new CPU, so if you don't have an old CPU at hand, it's going to be a fuss having to request a chip to do the update. That's why I would generally stick to the motherboard the CPU was released with, i.e first generation with B350/X370, second generation with B450/X470, etc. You don't need to buy a more expensive motherboard to use with an older CPU because it's not necessary and vice versa, you don't need an older CPU to update the BIOS before it takes the new CPU.

[comment deleted by staff]
  • 1 month ago
  • 1 point

Underwhelming is the wrong word - meeting expectations is better. AMD are showing consistent improvements - gen 1 -> gen2 -> gen3. We were expecting 10-20% improvements across the board and AMD appear to be on target. Hardly a shocker or newsworthy. However Intel have been riding on Skylake far too long now, nearly 4 years. AMD now all caught up and cheaper. A strategic f%%% up by Intel, never let your competitor get within inches of the finishing line when you were several yards ahead. They are now under a microscope and leadership may relay urgency to the R&D teams. I hope they do not rush a chipset to launch. Kabylake was pointless, Coffee Lake should have been Coffee Lake Refresh as the entirely new launch.

  • 1 month ago
  • 1 point

To me it was underwhelming. Like I said, I was expecting a 10% increase in clock speed, which is very reasonable, and not as ridiculous as people expecting 5.0 GHz. The 2700X had an all-core boost of 4.0 GHz out of the box, is it so bad to expect 4.4 GHz all-core from the 3700X given the die shrink and we're now down to 7nm from 12nm? I mean, out of the box, it's only 4.4 GHz single-core boost, which is a measly 100 MHz from the 2700X's 4.3 GHz. That is just underwhelming for me. And then core and thread count, I mean, like I said, 4 cores, 8 threads is nearing the end of its days for gaming, to still release them now is a bit meh. 6 cores, 12 threads would've made more sense and with AMD's goal to take as much market share as possible with Intel falling severely behind, that would've been the deal breaker. As it stands though, it's still better value, but it leaves wriggle room for Intel.

I do agree Intel got far too comfortable. 10nm was so hyped two years ago and they failed to launch it. They clearly hit an obstacle in development. If they scrapped that and started from scratch, they'd be up to speed by now, but instead, they tried to push forward with it. It's too late to scrap now after so many years of development cost, but even pushing it out, it's going to be a tremendous loss, as they plan to release 7nm next year and 5nm the year after, according to a road map I saw before. Very very shaky, I wonder if they'll share how much money they lost at 10nm......

  • 1 month ago
  • 2 points

I was expecting a 10% increase in clock speed, which is very reasonable

Two things to note though:

  1. Clock speed isn't everything. Low IPC+High clock speed can still be worse than lower clock speed but high IPC. The FX series chips could hit 5GHz. A 2500k OC'd to 4.5GHz doesn't perform nearly as well as a 7600k stock, which only turbo boosts to 4.2GHz. My point is, substantial IPC improvement plus a bump in clock speed is likely more important than just hitting 5GHz. If AMD's numbers and demonstrations are to be believed, then stock 3800x should be a good match for the stock 9900k despite the lower boost clock. I'd wait for independent benchmarks to verify, but this is quite promising.

  2. Smaller processes can make higher clock speeds more difficult, not less. Check out this article for an overview of some of the challenges. I wouldn't be surprised at all if we didn't see 5GHz potential out of Intel's first Ice Lake chips when they finally release for similar reasons.

  • 1 month ago
  • 1 point

I know, it's just the increase is 100 MHz as it stands right now, that's... Nothing? It's like they didn't work towards it at all. 15% IPC improvement is quite big, that is more than I expected, without a doubt, so it seems like they focused purely on IPC improvements for third generation, which I suppose makes sense given part of their lacking compared to Intel was in this area. It's like you say though, need to wait for benchmarks to determine this.

Well, I'm no engineer, and like I said, I didn't expect an unreasonable increase like many who were expecting 5.0 GHz. Now that was laughable. There was just no way 5.0 GHz was going to be announced, even if they remained at 12nm and didn't drop to 7nm, I don't think AMD can pull off 5.0 GHz with their Ryzen CPUs.

  • 1 month ago
  • 1 point

Yeah, I think 5GHz is going to be increasingly difficult/rare, and that includes Intel going forward.

My point is the clock speed increase might look like nothing on paper, but the actual performance per core increase (or the numbers that really matter) are substantially better. 4.3 vs 4.4 is an unfair comparison, just like comparing a 4.5GHz 2500k to a 4.3GHz Ryzen 2700x is unfair. The actual throughput is very different since IPC is so much higher. Lazy math shows that the new chip would be equivalent to the old chip running at about 5GHz. Of course, need to wait for more benchmarks than a curated demo of Cinebench and PUBG before we really know how it performs. Still worth getting excited about though in my opinion.

  • 1 month ago
  • 1 point

One thing to bear in mind -> Previous generation Ryzen CPUs showed best value when you purchased the non XFR variants because generally they would overclock just as high as the XFR CPUs, making them pretty redundant to anyone who knew what they were buying. I find it a bit interesting that AMD has set up the Ryzen 7 lineup to only feature these XFR chips, but the upcoming Ryzen 5 linup includes both the 3600 and the 3600X.

  • 1 month ago
  • 1 point

I think you mean non-X variants. The X doesn't stand for XFR. 2600, 2700, along with first generation Ryzens all supported XFR, just the X variants were "factory overclocked" and thus had higher clock speeds out of the box. And what you're saying varied depending on price. I remember at one point less than a year ago, the 2700 and 2700X only had a $30 difference, making 2700X the better deal, frankly because if you wanted to overclock the 2700 to 2700X performance, you needed a pretty decent cooler which costed more than $30. Not long ago though, this was a bigger gap, when the 2700X was still selling $310 (not any more, fortunately) and the 2700 was selling at $230, then yes, the 2700 was the better deal.

Similarly, when the 2600X and 2600 had a $50 gap, it was debatable, both were close in value, since you needed a $50 cooler at least, like the Mugen 5/FUMA, to overclock the 2600 to 2600X performance. Strictly speaking, the 2600 was better value, because you'd be using a better cooler, but not everyone wants to work with overclocking. Then the price gap grew and 2600 was just superior in value, but about a month or two ago, there was only a $15 gap, so the 2600X was just far superior in value. Value shifts according to the price, and the price difference between the X and non-X CPUs have fluctuated quite a lot.

I wouldn't be set on the line ups yet, there's still E3 and I'm pretty sure AMD has more CPUs to announce, so there might be non-X variants of all the X CPUs right now. Keep in mind we haven't gotten word on Ryzen 3 CPUs and the apparently confirmed and seen at Computex 16 core, 32 thread CPUs, meaning it's very possible that non-X variants could exist and just not announced yet. Due to the binning process, I can't imagine there not being non-X variants of them all, because AMD would have to throw away a lot more chips if that was the case due to failure to meet the specs, so I'm almost certain there will be lower binned CPUs to keep these for sale.

  • 1 month ago
  • 1 point

More likely that AMD has thown off the "underdog" label and now expects to price themselves much closer to Intel's competition? They have the power, people are learning that AMD chips have said power, and now AMD wants (more of) the pricetag that goes along with it?

Didn't they recently brag in an investor's meeting that their margins were increasing (presumably from after the post-cryptoboom crash)?

The low prices on state of the art AMD chips is too good to last. But there still are some outrageous deals on first and second generation zens to be had.

  • 1 month ago
  • 1 point

I'm not complaining about price compared to Intel, because the 3700X is a deal, compared to the 9900K, regardless. Like I said, I'm not optimistic that it will beat the 9900K overclocked to 5.0 GHz, but my guess is that it'll only be 10% slower at most (pure random guessing) so being $329 vs $490 (current market price of the 9900K) it is a bargain still.

My issue with the 3800X is comparing it to the 3700X, so hopefully there are going to be surprises from the extra $70 or it's going to be more stupid than the 8500 and 8600 from Intel that were higher binned versions of the 8400.

No idea, I don't really keep up with news like that, but I do know AMD's revenue is nowhere Intel's, which makes it more surprising that they're not busting out Ryzen 3s as 6 cores, 12 threads to absolutely steal market share for future revenue.

You're right on that, some first and generation CPUs are currently pretty good deals. The 1700/1700X have been worthy deals for a while now, the 2600X came down to something like $160 before, although I think it's gone back up slightly now, but still worth a look, and I expect the 2700X price to come down a bit more soon. Gotta wait for benchmarks, but I will guess when third generation comes out, second generation is very likely to be a good deal, similar to when second generation came out, first generation was a splendid deal, where the 1700X went for as cheap as $150 during Black Friday.

  • 1 month ago
  • 1 point

AMD hype train in effect. Slower performance but MOAR COREZ!!111!11 at maybe 25% of the price of Intel. Big deal.

  • 1 month ago
  • 2 points

why do you think it wil be slower

  • 1 month ago
  • 1 point

Last time I checked 15% IPC improvement is not 'slower'.

  • 1 month ago
  • 2 points

AMD is claiming 15%.

Intel is claiming 18%.

AMD is increasing core counts to 12.

Intel is increasing core counts to 10.

Not a whole lot is actually changing in how either market or in target markets for both.

  • 1 month ago
  • 1 point

Intel is claiming 18% IPC improvement on a chip that will be limited to 4.1GHz (10nm), giving it at best the performance of a 4.7GHz chip. AMD's claims can be even harder to justify (they claimed and IPC boost from zen to zen+ that required including faster clocks to justify), but may have the performance in most cases.

Expect the "Ryzen 3000 APU" to be a bigger part of AMD's strategy, if not part of the hypetrain. Breaking into the desktop market makes a lot of noise and shows up in places like this. Breaking into the laptop market shows up on both Main St. and Wall St. Rome seems to be doing the same thing as the zen1 Epyc, only better. If it goes mostly like promised, expect plenty of places that bought zen1 Epyc for testing to buy zen2 [Rome] Epyc for production servers.

  • 1 month ago
  • 2 points

Intel is claiming 18% IPC improvement on a chip that will be limited to 4.1GHz (10nm), giving it at best the performance of a 4.7GHz chip.

For a mobile chip that is a huge improvement.

Expect the "Ryzen 3000 APU" to be a bigger part of AMD's strategy, if not part of the hypetrain.

AMD has already confirmed the APU are refreshes on 12nm not new 7nm silicon.

  • 1 month ago
  • 1 point

I still expect a Zen 2 APU roughly next year. The first APUs were branded "2200" even though they used zen[no plus], and were roughly a year later than the Zen CPUs. No clue if they will replace the I/O chip or try to make an entire SOC out of half a chiplet. Having one CPU chiplet and one GPU chiplet makes too much sense, perhaps eventually they will make one.

Sort

add arrow-down arrow-left arrow-right arrow-up authorcheckmark clipboard combo comment delete discord dots drag-handle dropdown-arrow errorfacebook history inbox instagram issuelink lock markup-bbcode markup-html markup-pcpp markup-cyclingbuilder markup-plain-text markup-reddit menu pin radio-button save search settings share star-empty star-full star-half switch successtag twitch twitter user warningwattage weight youtube