>AMD's low budget APUs beat Intel's i5s Why do

>3dMark
OP, you could at least try

>>56450626
A12s are their top end apus.

>>56450626
>believing AMD branded benchmarks

You also don't understand what APU means

AMD's APU's have always bean better GPU wise.

>>56450693
all new apus are low budget for am4 platform

File: Screenshot_2.png (75KB, 199x163px) Image search: [Google]

75KB, 199x163px

>>56450738
>A12s are their top end apus.
>top end apus.
>APU

File: 1456993379781.jpg (130KB, 840x473px) Image search: [Google]

130KB, 840x473px

This new APU's look quite nice, the top end 9800E has a 35W TDP with 4 modules and 3.1 Ghz base clock, my 7650K has a 3.3 Ghz base clock with 65W at the lowest cTDP, both do the same boost 3.8 Ghz
Unfortunately I can't see anyone but OEM's interested in them, Zen APU's are around a year from now, and Zen is around 6 months from now
Maybe they would be nice as a placeholder for an AM4 mobo if you're desperate to get something new and still want to wait for Zen

>>56450626
I hope this is true

>>56450765
AM4 not FM2 on AM4 we will get ZEN APU

7th gen amd beating 6th gen intel core, really make u think, got a new 7 gen intel core benchmark?

>>56450990
AMD BTFO!!!

>>56451006
> The AMD PRO A-Series APU combines advanced technologies on a single chip for a revolutionary computing experience: exceptional graphics, up to 12 compute cores (4 CPU + 8 GPU)2, and a highly responsive system. Our latest APUs deliver over 950 gigaflops of compute power. That’s serious PRO-level processing for seamless performance, online collaboration, and productivity.

- http://www.amd.com/en-us/solutions/pro/performance

https://en.wikipedia.org/wiki/TOP500#/media/File:Supercomputers-history.svg

xD

>>56450626
Because most aren't poor and can afford a dGPU.

>>56451703
You do know that 80% of computers have I GPUs right

File: NVIDIA-GeForce-GTX-1060-Performance-and-Efficiency-Benchmarks.jpg (73KB, 1200x622px) Image search: [Google]

73KB, 1200x622px

It's mostly graphic workloads but thank fuck the graphs go from 0% like a normal graph should.

AMD CPUs work well for synthetic because of the way bulldozer-like architectures are structured

>>56450785
>Unfortunately I can't see anyone but OEM's interested in them
Starting prices for Zen will be between $300-600, so I can totally see who who be interested in them: poorfags

>>56452510

>Performance
>RX 480 1
>GTX 1060 1.2

Anon, seriously?

>>56452843
Considering 80% of Intel's range also falls in that $300-$600 price range, by your own 'logic' that means people who buy Intel are also poor.

>>56452884
>that means people who buy Intel are also poor.
If they buy i3s and lower, then yes.

>>56450785
>>56450626
>A12-9500 3.8/4.2 GHz
>i5-6500 3.2/3.6 GHz
>same number of cores, but needs to be clocked higher to match the i5's multithreaded performance
It's still shit. Call me when they can beat an i5-6600K at 4.2GHz

>>56452893
you said $300-$600 range
i5 6600K is $234 on jewmazon, so it's not the cheapest one that can be found

how is that price range for poorfags

>>56453291
Zen will start at $300-600. I don't care about the i5-6600k because that's the CPU poorfags buy when they want to live on ramen noodles for a month to afford it.

Their APUs still only have 8 CUs, which means they are still shit, who cares.

>>56450626
>3dmark
An APU has better graphics performance than a CPU with integrated video

In what world is this supposed to be impressive?

>>56452876
>graph goes from .8 to 1.5
It's pretty fucking retarded and skews the graphs to look like Nvidia's 2-3x faster when they're barely 15%.

>>56453019
It's really just impressive that they've taken Bulldozer and reworked it into this.

Bulldozer would've needed twice as high of clocks as that 6500 to reach the same performance and at 3x the power draw.

This still needs higher clocks, but manages to do it at nearly the same power draw.

>>56450626
i3 and i5 considered "performance", any year after 2012 is the real meme

>>56452510
That would make graph less readable.

>>56453479
>An APU has better graphics performance than another APU

File: 30a67041287e2969d4b8172dbf329d82.jpg (139KB, 752x1062px) Image search: [Google]

139KB, 752x1062px

>>56450626
>only has 8 PCIe lanes from the APU
>only has 6 PCIe lanes (and not even Gen 3 like Lynx Pont) from the highest end chipset AT BEST
AhahahaAHAHAHAHAHAHAHAHAHAHAHAHA

>>56453550
^this

My i5 6600 is very aggressive in power savings.

>>56453954
What's the point of having more PCIe lanes?

>>56453786
Except Intel's CPUs aren't being sold as APUs. The integrated video is a bonus.

For example, I just bought an Intel i5-6600k with a dedicated graphics card. Am I ever going to use the integrated graphics? No. Did I spend a little extra money for a graphics chip I'm not going to use? Of course, I got jewed. Intel is a corporation after all.

However, I would NEVER spend a premium on an AMD APU if I was planning on using a dedicated graphics card. So they are not comparable at all.

>>56454201
Can't use an M.2 drive at its fullest speed, can't use 10GbE NICs to its max bandwidth, can't use two graphics cards, etc.
Besides, this REALLY doesn't look good for Zen's competitiveness in the HEDT market.

>>56454201
M.2, multiple gpu's, expansion cards, usb.

>>56454231
>sold as
"marketed as" is more appropriate

>>56454231
Would you use an Athlon X4 950 knowing that it only has half the available bandwidth for any graphics card?

>>56454231
You could just buy the Athlon X4 950 if you don't need an iGPU.

>>56450785
So what's the difference between the Athlon X4 845 and the X4 950? Other than the DIMM standard and the socket type?

>>56453572
how does it make is less readable than literally skewing information?

>>56453019
What's impressive it's that it matches the 6500 at almost the same power usage, with a 28nm node despite having somewhat highish clocks
They really polished the turd that was the shit 28nm GloFo process they're using, and the turd that was bulldozer, it even feels like Bulldozer finally became what it was supposed to be
>>56454231
>Except Intel's CPUs aren't being sold as APUs. The integrated video is a bonus.
It's not, there's i7's and Xeon's that lack a iGPU for people that don't need a iGPU
>>56454232
First gen Zen isn't going head to head against Intel's HEDT, at best it will be competitive with most i7's, FPU wise Intel will still keep absolute dominance
They're heavily targeting the server market with Zen, Zen+ might compete head to head with Intel's HEDT though, a lot of stuff didn't make it to Zen due to area and energy constraints
All the chipsets launched today are for the mainstream market, there's a enthusiast chipset TBA that will be beefier
>>56454326
The X4 850 is Steamroller, the X4 950 is Excavator , in theory Excavator should have higher IPC and energy efficiency, but we won't really now until we get benchmarks
Also, some chipset improvements like USB 3.1 and NVMe

File: 1446840517980.jpg (160KB, 1024x576px) Image search: [Google]

160KB, 1024x576px

>>56454578
>All the chipsets launched today are for the mainstream market, there's a enthusiast chipset TBA that will be beefier

>>56454578
>It's not, there's i7's and Xeon's that lack a iGPU for people that don't need a iGPU
But these are generally more expensive. If you're in that price range, yeah, they're more cost-effective since you're getting a super powerful CPU and not wasting money on an iGPU you won't use.

In the lower price ranges however, where you typically get more bang for your buck, you're basically forced to waste money on an iGPU you're not going to use.

>>56454578
I said X4 845, which IS Excavator for the FM2+ socket

>>56452050
80% of computers are also in schools, offices, and other lines of work where dgpus just take more power and arent necessary.

>>56454232
>Can't use an M.2 drive at its fullest speed
Arguable, I can't find any sources on what version of PCIe would be best, but from what I see on various M.2 drive specs, one 3.0 lane would be enough to let it saturate.

>can't use 10GbE NICs to its max bandwidth
So where are you getting your 1.25GB/s internet from?

>can't use two graphics cards
Blatantly false, there's no difference between 4x3.0 lanes and 8x 3.0 lanes.
Chipset+SoC combined gives 8x3.0 lanes and 6x2.0 lanes.

4x2.0 go to the M.2 (could also just go with 2x2.0)
2x2.0 go to another PCI slot
4x3.0 go to 2 other PCI slots, or if one is unoccupied, run one at x8.

There you have full speed M.2, two full speed GPUs, and if manufacturers would put in an x4 slot instead of M.2 x4 you could run your 10GbE NIC at full speed instead of the 1GB/s of 2x2.0.

>>56454326
We don't know whether the 950 will have L3 cache but if it does, it'll mop the floor with the 845.

>>56454789
>Arguable, I can't find any sources on what version of PCIe would be best
I've tested by SM951 on an adapter card using lanes from the southbridge on both the Z97 and Z170. The Z97 is undoubtedly slower by at least 80% with increasingly smaller packet sized. Even though it doesn't seem to use all four lanes of PCIe 3.0, that's largely down to the controller on the drive and not the bandwidth made available to it. It will use as much of the bandwidth as possible.
>1.25GB/s internet
LAN connection. I have a NAS box that's 10GbE capable and a costly switch with two SFP connectors.
>Blatantly false
Disabled on the released chipsets, as stated by AMD. Even though I know AMD will allow Crossfire on their boards at some point, Nvidia cards won't work in SLI, and SLI is far more important to the current multi-GPU market due to Nvidia's market dominance.

>>56454927
Also, putting a M.2/PCIe SSD on the PCIe lanes directly from the CPU makes it go a smidge faster in almost every metric, especially with the Z97 boards. I'm guessing that might have to do with latencies from going to the southbridge through the DMI to the CPU rather than going to the CPU directly.

File: apu.png (39KB, 500x500px) Image search: [Google]

39KB, 500x500px

>low budget APU

>>56450990
Kaby Lake is just Skylake with higher clocks.

>>56455131
>Kaby Lake is just Skylake with higher clock
Oh, okay then. It's not like higher clocks are going to have any impact on a benchmark test, right?

>>56450626
that's for integrated graphics performance, holy shit

>>56455215
Nobody claimed otherwise :^)

>>56455215
PCMark has CPU-bound benchmarks, though it seems biased towards GPU's nowadays

I can't wait for Zen to come out.
I will order the highest end AMD Zen on day 0.

>>56455286

What I really hope for (though I doubt it will happen) is Zen - or at least the high performance versions - to be soldered. Plus i'd like AM3+ mounting to fit AM4. Plus i'd like a unicorn.

>>56453530
>1060
>faster
good joke

>>56452843
>Starting prices for Zen will be between $300-600
source?

What was your first AMD CPU, /g/?

Looks like I'll be getting an A12-9800 for a quiet workstation build

>>56455626
I never made an AMD build for myself. But my brother had a low budget and I built him a pretty decent gaming rig with an AMD Phenom II X4 955 black edition. It got the job done.

>>56455626
A first gen Athlon that ran at 600 Mhz paired with a Trident GPU

>>56455007
I also have an sm951 nvme drive and I think it throttles more than saturates. That said new drives and interfaces are coming out quick pci-e x8 I believe it is.

>>56455626
AMD 386 @ 40 MHz

>>56455783
No, it's definitely the controller on the SSD. There is no way that the SM951 is pushing the current M.2 standard with its current simultaneous read/write speeds. It should be way faster than that. Either the controller still can't push past PCIe 2.0 (not 3.0) 4x speeds, or the accessing the NANDs are causing the hold-back.

No new APUs with k, all of them has locked multipliers, what a bummer. Max base of 3.8 and max boost of 4.2 for 2 Excavator modules. Carrizo & BR confirmed to be non-overclockers for shit at that 28nm process node.

14nm LPP will hit 4GHz for 8 core zen on OC maybe? I'm not hopeful. 2.8-3.8 range looks more realistic for base-boost

>>56455830
>14nm LPP will hit 4GHz for 8 core zen on OC maybe
Nope. It looks like it will struggle to break 3.3GHz, if the sample chips are anything to go by.
RIP Zen

The real question is how will a A12-9800 compare to a Pentium G4400 + Nvidia 730GT build in proprietary video games?

>>56456010
>A12-9800 compare to a Pentium G4400
Not well, since single-threaded games still abound and Excavator is still at least 20% behind Skylake in that instance.

>>56450765
>APU my son poo in loo or poo in street?

>>56456010
>Nvidia 730GT
why

>>56456010
The GPU in an A12-9800 would shit all over a crappy GT 730.

>>56456056
>The GPU in an A12-9800 would shit all over a crappy GT 730
No it won't. The GPU in the A12 struggles to match the APU in the Xbone. A GT 730 can at least match the XBone's performance.

>>56453358
>6600k is on par with an I7
>bad cpu
even the 6500 is as good as an i7 for gaming if gta v is anything to go by since it relies alot on the cpu

>>56456085

>A GT 730 can at least match the XBone's performance.

It really doesn't.

>>56456213
It does if you lower the settings down to the XBone's level. Nvidia just works better on PCs.

File: perfrel_1920_1080.png (41KB, 500x1130px) Image search: [Google]

41KB, 500x1130px

>>56456247

I don't think you understand the level of potato the 730 is. The closest desktop equivalent to the chip found in the xbone is the same chip in the 370 and the 370 fights a 760 - a far cry from the potato of the GT 730 (it doesn't even have GTX branding).

>>56456307
>370 fights a 760
Don't make me laugh. A GT 740 can beat the R7 370. The Xbone is weaker than the 370 anyway.

File: killer instinct 1080p.jpg (134KB, 548x629px) Image search: [Google]

134KB, 548x629px

>>56456327

I guess Nvidiocy really is a condition on /g/. Notice the 7850 (aka 370 give or take) miles ahead of chips that are faster than the 730. Especially considering the 7850 isn't factory clocked as high as a 370.

>>56456366
>KI
>not AMD-"optimized" garbage

>>56456402

>i-i-i-t doesn't count.

Its not like you could actually provide any evidence showing the 739 faster than a 785/whatever it was rebranded into (and it had a lot).

File: goyforce.jpg (64KB, 335x478px) Image search: [Google]

64KB, 335x478px

>>56456327
>GT 740 can beat the R7 370
GT 740 specs:
384 shaders @ 993MHz with DDR3
R7 370 specs:
1024 shaders @ 975 MHZ with GDDR5

Oh man which one better, so hard to guess.

>>56456755
GT 740 of course!

>>56456327
>A GT 740 can beat the R7 370
Are Nvidia fanboys this deluded? That crapfest card is only used by OEM's on shitty poorfag prebuilds for the sole purpose of having the Nvidia branding
I wonder how many Nvidiots are like this
>>56456755
Aren't there GDDR5 models of the GT 740? With Nvidia's fuckyou branding on the low end you never know, but DDR3 would seriously bottleneck most GPU's

File: 1795t4.jpg (81KB, 500x616px) Image search: [Google]

81KB, 500x616px

>>56457062

>I wonder how many Nvidiots are like this

All of them.

>>56450626

Why is it okay to post benchmarks of AMD hardware by AMD yet if you do the same for nvidia you're a shill and it's totally fake?

>>56453358
>im so edgy right now
>look at how much attention im getting
>all the grils will like me at school now
>reward myself with chicken tendies and anime and also emo musix!

>>56457292
Because one of these two companies has a long-standing record of lying or misrepresenting their products.

>>56457441

HAHAHAAH

AMD has basically trademarked the term "up to" to scam their benchmark results.

>>56457441
So quite literally it's okay when AMD does it.

Nice to know, Pajeet.

>>56457465
No, the point is that AMD claims tend to be significantly more reliable than Nvidia's. Take the 3.5GB fiasco for example. Name one event where AMD did something of this magnitude.

File: rop989l.jpg (603KB, 1920x1440px) Image search: [Google]

603KB, 1920x1440px

>>56457493

Claiming their dual GPU card had twice as much usable RAM as it acutally does.

But as well all know, it's okay when AMD does it and you will defend this.

>>56457726
Oh, you're clinically retarded.
Wow, this changes everything!

>>56457726
Every Dual GPU ever said that

>>56453954
sucks for the 0 people who run APU setups with m.2 SSDs in them

>>56456085
So you think a 730 matches a 7790?

File: TEGRA STRONK.jpg (91KB, 680x384px) Image search: [Google]

91KB, 680x384px

>>56457292
Because AMD doesn't do things like pic related, where Tegra apparently scales to infinity because NVidia chose a benchmark that was specifically made for their shit and won't run on anything else.

Would be nice if AMD actually managed to do nice things again like 15 years ago.
Intel having no competition only allows for us being jewed out and tech stagnation.

>>56456327
The Xone GPU is a modded 7770, that performs almost the same as a GTX750ti

>>56457292
This slides are literally the only existing benchmarks for the new APU's, maybe in a few days we will see benchmarks from other sources

>>56456022
poo in the GPU, father

>next upcoming amd will beat nvidia/intel!!!!!

It comes out and sucks dick and worse compared to nv/int

>next upcoming amd product will beat nvidia/intel!!!

Repeat cycle

>>56457062
I have a couple GT 730's w/ GDDR5 not DDR3 and they're great little cards if you pick them up on a sale or as a stopgap. No way a 730 shits on a 370 ever. I even still have a 7950 and a 290 running. Nvidia likely has GDDR5 versions of the 740 as it does of the 730 (I own them). I will say the 730 is ever so slightly better than the R7 240 I have. R8 250 is better I wager.

>>56457916
>Born in the 90's
15 years ago? How old are you? Intel = good is recent thing.

When are the AM4 motherboards gonna get released?

>>56462177
When american mega dicks decides to release em

>>56450626
>6500
>non K

>>56453019
It's only using 17% higher clock rates with the same tdp on a node double the size. With a vastly more powerful gpu on top of all that. This is what zen is supposed to have 40% higher IPC than.

>>56455626
900mhz duron. Thing was a beast that refused to quit. Think I still have the mobo and cpu in a box somewhere.

>>56462372
You do realize that the benchmark in question is mostly GPU-related rather than pure CPU benchmark?

Performance as a radiator? xd

>>56462478
You do realize that the entire point of an apu is to actually accelerate processing power via the integrated gpu, right? Shame nothing actually utilizes it that way, though.

>>56462565
To add to this, if we look at it objectively on the graphical standpoint. WIth 8 CUs the iGPU will be just over half of an rx 460. Given that knowledge, the performance would be somewhere around a GT 740, maybe leaning towards a 750.

perf/watt > raw perf when it comes to a web browsing machine/general purpose laptop

AMD can achieve slightly better performance than Intel's GPUs, but requires over 2x the power consumption to do it.

>>56462924
That isn't even remotely true. Intel's IGP arch is terrible, and their chips grossly exceed TDP under heavy GPU load.

>>56462949

[citation needed]

File: 5Oh3qMG_.jpg (12KB, 300x300px) Image search: [Google]

12KB, 300x300px

TLDR:
A circlejerk for poorfags obtaining gratification from buying cheap shitty hardware
>not buying GTX 1070, i7 6700K for minimal 2016 tier performance

>>56463024
little money more brain is better than more money little brain bby

>>56462963
>hurrrrrrrrrrrrrrrrrrrrrrrrrrrrrf give me a citation for this simple fact that anyone who actually read reviews would know
>BUT I DON'T HAVE TO PROVIDE A SOURCE FOR MY BASELESS SHILLING

Core M all the way to their 45w HQ SKUs with Iris Pro graphics, the chips suck down power like a motherfucker when the IGP is stressed. It has been this way for years. Not knowing this instantly brands you as a tech illiterate retard or an autistic fanboy shill.
Intel's IGP arch is terrible when it comes to perf/watt. Their first mobile Iris Pro mobile chips were pulling 90w~ full load.

http://www.notebookcheck.net/Dell-Latitude-14-E7470-Ultrabook-Review.161557.0.html

>>56463029
>not having both
it's like you've chosen to be a submissive boipussi

Given that they're literally already beating Intel at the same power envelopes - at a lower cost -I think it's safe to say that Raven Ridge is going to utterly wreck shit.

>>56463080
indeed, i have neither little brain or little money, so I feel sorry for you man

File: 80059.png (18KB, 650x250px) Image search: [Google]

18KB, 650x250px

>>56463063

Read your own source, none of those notebooks exceeded TDP like you claimed.

>brands you as a tech illiterate retard or an autistic fanboy shill.

you tried to claim Intel has shit iGPUs and then BTFO'ed yourself and got reduced to fat fingering insults within one post.

just posting for a test ignore this

ps. I think intel has something up their sleeves not saying what

What's the ETA on this actually releasing? I need something better than the FX-6350 I have now. And the convenience of already having an AM4 board when Zen drops will be nice.

>>56463184

Again, read your own source. Notebookcheck's power consumption numbers include the power draw of every component in the system (including the screen etc) and don't concretely show anything about the TDP of the CPU or GPU.

>Intel's IGP arch is awful, theres a reason why intel isn't in the discrete GPU market.

Intel isn't in the dGPU market because their GPUs suck, they're not in it because there's no money in it. Gaming is dying and the HPC side of things is already covered by xeon phi (which doesn't have rasterization hardware and all the other legacy fixed function ASICs a GPU needs to be relevant for gaymen).

Getting angry and trying to be a keyboard warrior will not change the fact that Intel has superior offerings to AMD across the board currently. AMD should invest more in R&D instead of firing engineers and giving bonuses to executives.

File: Screenshot_20160906-233058.png (112KB, 540x960px) Image search: [Google]

112KB, 540x960px

>>56463184
Not so fast buddy I can't allow you to do that

>>56463224
>tech illiterate retard doesn't understand what hes even saying
Its called a power delta, you 70 IQ dipshit.
The screen doesn't pull more power when the CPU is under load. The only increased power draw is from the SoC. The power delta idle vs load shows what the CPU+GPU/SoC are pulling.
The chips are grossly exceeding TDP when the IGP is stressed.

You're flat out lying like a little console fanboy on /v/

>>56463246
>Its called a power delta, you 70 IQ dipshit.
>The screen doesn't pull more power when the CPU is under load.

if you're going from idle to 100% load then the screen is also going from off or low brightness to max brightness. it's not uncommon at all for screens to pull >30w on their own. notebookcheck's power consumption numbers are only good for estimating battery life, NOT for drawing conclusions about power draw by individual components.

>>56463284
Oh look, the butthurt lying fanboy making more bullshit up.
No laptop on the market has high output dual CFL back lighting, child. The screen isn't pulling 30w.

You're grasping at straws, desperately attacking the source because it proves you wrong. Screen power is accounted for, there is no dramatic shift in brightness level, the LED backlight in a tiny ultrabook isn't drawing even 5w. The power delta is showing what the SoC is pulling.

>>56463335
>No laptop on the market has high output dual CFL back lighting, child. The screen isn't pulling 30w.

huh? CCFL screens would use way more than 30w. 30w for an LED lit screen is extremely common.

>You're grasping at straws, desperately attacking the source because it proves you wrong.

how have i attacked the source? all i've done is show that it isn't corroborating your claims, not that it isn't credible or some other BS.

> Screen power is accounted for,

[citation needed].

>>56463368
>30w for an ultrabook laptop screen is common!
>BUT YET AGAIN I DONT NEED TO PROVIDE SOURCE FOR MY OWN SHILL CLAIMS


This is what a desperate shill looks like

>>56463384

as expected you resorted to pure adhom/insults from simple questions about your claims, lul.

better luck next time friend :-)

>>56463414
Yep, desperate shill shitposting.
Trying to take the high ground when confronted about his absurdist claims.

Good for them but my A8 gets raped by my broadwell-u iris 6100

>>56463024
So you're gonna buy it for me then. Thanks anon, you're a champ.

>>56450626
>Why do so many people still fall for the Intel meme?
because I can get old xeons much easier than old opterons
and they're better as well

>>56452893
w/e poorfag

I'm using power9 only

enjoy microshaft wangblows you fucking faggot idiot retard nigger autist

>>56457851
>>56454240
>>56454232
>m.2
I'm still using ancient tier sata 3
I don't have thousands of dollars to waste on ludicrously fast and expensive SSD's and NIC's

>>56463024
>i7
literally a i5 with hyperthreading
I don't know how that's worth $100 or more